Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- H.-J. ZhangSuggested from co-authorship
- Hong-Jiang ZhangSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works559 from public data
- Counterfactual VQA: A Cause-Effect Look at Language Bias565
A novel counterfactual inference framework is proposed, which enables the language bias to be captured as the direct causal effect of questions on answers and reduced by subtracting the direct language effect from the total causal effect.
- Spatio-Temporal AutoEncoder for Video Anomaly Detection563
A novel model called Spatio-Temporal AutoEncoding (ST AutoEncoder or STAE), which utilizes deep neural networks to learn video representation automatically and extracts features from both spatial and temporal dimensions by performing 3-dimensional convolutions, which enhances the motion feature learning in videos.
- A generic framework of user attention model and its application in video summarization553
A generic framework of a user attention model is presented, which estimates the attentions viewers may pay to video contents, and a set of modeling methods for visual and aural attentions are proposed.
- Correlative multi-label video annotation504
A third paradigm is proposed which simultaneously classifies concepts and models correlations between them in a single step by using a novel Correlative Multi-Label (CML) framework and is compared with the state-of-the-art approaches in the first and second paradigms on the widely used TRECVID data set.
- Tag ranking496
This paper proposes a tag ranking scheme, aiming to automatically rank the tags associated with a given image according to their relevance to the image content, and applies tag ranking into three applications: tag-based image search, tag recommendation, and group recommendation.
- 458
- Quantization Networks432
This paper provides a simple and uniform way for weights and activations quantization by formulating it as a differentiable non-linear function that will shed new lights on the interpretation of neural network quantization.
- Towards Federated Learning in UAV-Enabled Internet of Vehicles: A Multi-Dimensional Contract-Matching Approach318
This work proposes the adoption of a Federated Learning (FL) based approach to enable privacy-preserving collaborative Machine Learning across a federation of independent DaaS providers for the development of IoV applications, e.g., for traffic prediction and car park occupancy management.
- Feature Pyramid Transformer317
This work proposes a fully active feature interaction across both space and scales, called Feature Pyramid Transformer (FPT), which transforms any feature pyramid into another feature pyramid of the same size but with richer contexts, by using three specially designed transformers in self-level, top-down, and bottom-up interaction fashion.
- A Survey on Deep Hashing Methods278
This survey detailedly investigates current deep hashing algorithms including deep supervised hashing and deep unsupervised hashing, and categorizes deep supervised hash methods into pairwise methods, ranking-based methods, pointwise methods as well as quantization according to how measuring the similarities of the learned hash codes.
- Camera-Aware Proxies for Unsupervised Person Re-Identification268
This paper proposes to split each single cluster into multiple proxies and each proxy represents the instances coming from the same camera, and designs both intra and inter-camera contrastive learning components for the Re-ID model to effectively learn the ID discrimination ability within and across cameras.
- HoMM: Higher-Order Moment Matching for Unsupervised Domain Adaptation268
This work proposes a Higher-order Moment Matching (HoMM) method, and further extends the HoMM into reproducing kernel Hilbert spaces (RKHS) and shows that the first-order HoMM is equivalent to Maximum Mean Discrepancy (MMD) and the second-orders are equivalent to Correlation Alignment (CORAL).
- 259
- 256
- Active learning in multimedia annotation and retrieval255
A survey on the efforts of leveraging active learning in multimedia annotation and retrieval, including semi-supervised learning, multilabel learning and multiple instance learning, focuses on two application domains: image/video annotation and content-based image retrieval.
- Counterfactual Zero-Shot and Open-Set Visual Recognition249
A novel counterfactual framework for both Zero-Shot Learning (ZSL) and Open-Set Recognition (OSR), whose common challenge is generalizing to the unseen-classes by only training on the seen-classes is presented.
- 248
- 246
- Class Re-Activation Maps for Weakly-Supervised Semantic Segmentation235
An embarrassingly simple yet surprisingly effective method: Reactivating the converged CAM with BCE by using softmax crossentropy loss (SCE), dubbed ReCAM, which not only generates high-quality masks, but also supports plug-and-play in any CAM variant with little overhead.
- Gradient Centralization: A New Optimization Technique for Deep Neural Networks228
It is shown that GC can regularize both the weight space and output feature space so that it can boost the generalization performance of DNNs, and improves the Lipschitzness of the loss function and its gradient so that the training process becomes more efficient and stable.
- Distilling Causal Effect of Data in Class-Incremental Learning218
The proposed causal effect distillation method can improve various state-of-the-art CIL methods by a large margin and capture the Incremental Momentum Effect of the data stream, which can help to retain the old effect overwhelmed by the new data effect, and thus alleviate the forgetting of the old class in testing.
- PLBP: An effective local binary patterns texture descriptor with pyramid representation212
The conventional local binary pattern is extended to pyramid transform domain (PLBP) by cascading the LBP information of hierarchical spatial pyramids, PLBP descriptors take texture resolution variations into account and show their effectiveness for texture representation.
- Cloth-Changing Person Re-identification from A Single Image with Gait Prediction and Regularization200
This paper introduces Gait recognition as an auxiliary task to drive the Image ReID model to learn cloth-agnostic representations by leveraging personal unique and cloth-independent gait information, and introduces this framework as GI-ReID, which enables an efficient and latency-free person identity matching for surveillance.
- 183
- Flickr distance183
Both subjective user study and objective evaluation show that Flickr distance is more coherent to human perception than Google distance, which makes it a more reasonable distance metric.
- 182
- Graph Contrastive Clustering175
This work first proposes a novel graph contrastive learning framework, and then applies it to the clustering task, resulting in the Graph Constrastive Clustering (GCC) method, which incorporates the latent category information to reduce the intra-clusters variance as well as increase the inter-cluster variance.
- 172
- Graph-based semi-supervised learning with multiple labels164
A novel graph-based learning framework in the setting of semi-supervised learning with multiple labels is proposed, characterized by simultaneously exploiting the inherent correlations among multiple labels and the label consistency over the graph.
- Bayesian video search reranking161
Two new methods are proposed in this paper to measure the ranking distance based on the disagreement in terms of pair-wise orders, which represents the disagreement between the objective ranking list and the initial text-based.
- 158
- 157
- 156
- Online video recommendation based on multimodal fusion and relevance feedback155
This paper presents a novel online video recommendation system based on multimodal fusion and relevance feedback, and is able to recommend videos without users' profiles.
- Joint Auction-Coalition Formation Framework for Communication-Efficient Federated Learning in UAV-Enabled Internet of Vehicles154
A joint auction-coalition formation framework to solve the allocation of UAV coalitions to groups of IoV components and it is shown that as the cooperation cost of the Uavs increases, the UAVs prefer to support the Iov components independently and not to form any coalition.
- Deep CTR Prediction in Display Advertising154
A novel deep neural network (DNN) based model is introduced that directly predicts the CTR of an image ad based on raw image pixels and other basic features in one step.
- A Survey of Efficient Reasoning for Large Reasoning Models: Language, Multimodality, and Beyond147
This survey provides a comprehensive overview of recent efforts aimed at improving reasoning efficiency in LRMs, with a particular focus on the unique challenges that arise in this new paradigm.
- 144
- Real-time large scale near-duplicate web video retrieval135
A novel method to address the efficiency and scalability issues for near-duplicate video retrieval by introducing a compact spatiotemporal feature to represent videos and constructing an efficient data structure to index the feature to achieve real-time retrieving performance.
- 132
- MaCAR: Urban Traffic Light Control via Active Multi-agent Communication and Action Rectification126
This work proposes a novel Multi-agent Communication and Action Rectification (MaCAR) framework which enables active communication between agents by considering the impact of synchronous actions of agents and outperforms state-of-the-art methods on both synthetic and real-world datasets.
- Visual query suggestion125
A new query suggestion scheme named Visual Query Suggestion (VQS) is proposed which is dedicated to image search and provides a more effective query interface to help users to precisely express their search intents by joint text and image suggestions.
- 122
- VideoSense120
A novel advertising system for online video service called VideoSense, which automatically associates the most relevant video ads with online videos and seamlessly inserts the ads at the most appropriate positions within each individual video, is presented.
- Image retagging116
This paper proposes a social image "retagging" scheme that aims at assigning images with better content descriptors and shows the remarkable performance improvements brought by retagging via two applications, i.e., tag-based search and automatic annotation.
- Contextual Video Recommendation by Multimodal Relevance and User Feedback114
VideoReach represents one of the first attempts toward contextual recommendation driven by video content and user click-through, without assuming a sufficient collection of user profiles available, and is designed to seamlessly integrate both multimodal relevance and user feedback by relevance feedback and attention fusion.
- 111
- Clickage111
It is argued that the massive amount of click data from commercial search engines provides a data set that is unique in the bridging of the semantic and intent gap, and preliminary studies on the power of large-scale click data are presented.
- Optimizing multi-graph learning111
It is shown that various crucial factors in video annotation, including multiple modalities, multiple distance metrics, and temporal consistency, all correspond to different correlations among samples, and hence they can be represented by different graphs, and OMG-SSL is able to simultaneously deal with these factors within a unified framework.
- Disentangled Representation Learning for Text-Video Retrieval110
This paper first studies the interaction paradigm in depth, where it is found that its computation can be split into two terms, the interaction contents at different granularity and the matching function to distinguish pairs with the same semantics.
- Active Boundary Loss for Semantic Segmentation110
Experimental results show that training with the active boundary loss can effectively improve the boundary F-score and mean Intersection-over-Union on challenging image and video object segmentation datasets.
- FECANet: Boosting Few-Shot Semantic Segmentation With Feature-Enhanced Context-Aware Network109
A feature enhancement module is proposed to suppress the matching noise caused by inter-class local similarity and enhance the intra-class relevance in the naive correlation, and a novel correlation reconstruction module is proposed that encodes extra correspondence relations between foreground and background and multi-scale context semantic features.
- 109
- Automatic location of text in video frames109
Experimental results show that the new text location scheme proposed in this paper is accurate and can be used to significantly reduce false alarms.
- An Adversarial Approach to Hard Triplet Generation107
This work proposes an adversarial network for Hard Triplet Generation (HTG) to optimize the network ability in distinguishing similar examples of different categories as well as grouping varied examples of the same categories.
- 107
- Online Convolutional Reparameterization104
This paper presents online convolutional re-parameterization (OREPA), a two-stage pipeline aiming to reduce the huge training overhead by squeezing the complex training-time block into a single convolution, and introduces a linear scaling layer for better optimizing the online blocks.
- 103
- Boosting Semantic Human Matting With Coarse Annotations102
This paper trains a mask prediction network to estimate the coarse semantic mask using the hybrid data, and proposes a quality unification network to unify the quality of the previous coarse mask outputs to boost end-to-end semantic human matting without trimaps as extra input.
- 102
- 102
- Tag refinement by regularized LDA96
A novel graphical model, regularized Latent Dirichlet Allocation (rLDA), is presented, which facilitates the topic modeling by exploiting both the statistics of tags and visual affinities of images in the corpus.
- NTIRE 2022 Challenge on Efficient Super-Resolution: Methods and Results94
The NTIRE 2022 challenge was to super-resolve an input image with a magnification factor of ×4 based on pairs of low and corresponding high resolution images and the aim was to design a network for single image super-resolution that achieved improvement of efficiency measured according to several metrics.
- 93
- Structural and Statistical Texture Knowledge Distillation for Semantic Segmentation92
This paper introduces a Contourlet Decomposition Module (CDM) that decomposes low-level features with iterative laplacian pyramid and directional filter bank to mine the structural texture knowledge and proposes a Denoised Texture Intensity Equalization Module (DTIEM) to adaptively extract and enhance statistical texture knowledge through heuristics iterative quantization and denoised operation.
- 92
- 91
- SLV: Spatial Likelihood Voting for Weakly Supervised Object Detection89
A spatial likelihood voting (SLV) module is proposed to converge the proposal localizing process without any bounding box annotations, and an end-to-end training framework for multi-task learning is proposed.
- 89
- 89
- DCT-Mask: Discrete Cosine Transform Mask Representation for Instance Segmentation88
This paper proposes a new mask representation by applying the discrete cosine transform(DCT) to encode the high-resolution binary grid mask into a compact vector, termed DCT-Mask, which could be easily integrated into most pixel-based instance segmentation methods.
- CrowdReranking88
This paper presents a new method for visual search reranking called CrowdReranking, which is characterized by mining relevant visual patterns from image search results of multiple search engines which are available on the Internet.
- Multi-Task Vehicle Detection With Region-of-Interest Voting87
This paper proposes a novel vehicle detection scheme based on multi-task deep convolutional neural networks (CNNs) and region-of-interest (RoI) voting that achieves superior performance in vehicle detection compared with other existing published works.
- Video2Shop: Exact Matching Clothes in Videos to Online Shopping Images86
A novel deep neural network is proposed to explore a new cross-domain task, Video2Shop, targeting for matching clothes appeared in videos to the exactly same items in online shops, which outperforms the state-of-the-art methods.
- Dense Learning based Semi-Supervised Object Detection85
This paper proposes a DenSe Learning (DSL) based anchor-free SSOD algorithm, and introduces several novel techniques, including an Adaptive Filtering strategy for assigning multi-level and accurate dense pixel-wise pseudo-labels, an Aggregated Teacher for producing stable and precise pseudo-Labels, and an uncertainty-consistency-regularization term among scales and shuffled patches for improving the generalization capability of the detector.
- MSRA-MM: Bridging Research and Industrial Societies for Multimedia Information Retrieval85
A dataset named Microsoft Research Asia Multimedia (MSRA-MM) is introduced that aims to en- courage research in multimedia information retrieval and the related areas and the performance of state-of-the-art industrial techniques can be evaluated accordingly.
- PastNet: Introducing Physical Inductive Biases for Spatio-temporal Video Prediction83
A novel approach called Physics-assisted Spatio-temporal Network (PastNet) for generating high-quality video prediction by incorporating a spectral convolution operator in the Fourier domain, which efficiently introduces inductive biases from the underlying physical laws.
- 83
- Global Versus Localized Generative Adversarial Nets83
The LGAN can be applied to train a robust classifier that prefers locally consistent classification decisions on the manifold, and the resultant regularizer is closely related with the Laplace-Beltrami operator.
- Interactive Image Search by Color Map82
This article presents an interactive image search system, image search by color map, which can be applied to, but not limited to, enhance text-based image search, and proposes a simple but effective scheme to mine the latent search intention from the user’s input, and exploit the dominant color filter strategy to make the system more efficient.
- Video search re-ranking via multi-graph propagation81
Experimental results verify the effectiveness of the graph-based propagation approach combined with the object-sensitive query analysis approach, which brings significant improvement to the baseline of text-based video search.
- 81
- 79
- Social Media Retrieval79
This comprehensive text/reference examines in depth the synergy between multimedia content analysis, personalization, and next-generation networking and demonstrates how this integration can result in robust, personalized services that provide users with an improved multimedia-centric quality of experience.
- Semi-supervised kernel density estimation for video annotation79
An improved algorithm named semi-supervised adaptive kernel density estimation (SSAKDE) which employs local adaptive kernels rather than a fixed kernel, such that broader kernels can be applied in the regions with low density, and more accurate density estimates can be obtained.
- Dynamic Spatio-Temporal Graph-Based CNNs for Traffic Flow Prediction78
DST-GCNN is a two stream network that learns expressive features to represent spatio-temporal structures and predicts future traffic flows from surveillance video data and achieves competitive performances compared with the other state-of-the-art methods.
- Spatio-Temporal Inception Graph Convolutional Networks for Skeleton-Based Action Recognition78
This work designs a simple and highly modularized graph convolutional network architecture for skeleton-based action recognition by repeating a building block that aggregates multi-granularity information from both the spatial and temporal paths.
- Finding image exemplars using fast sparse affinity propagation78
A novel approach to organize image search results obtained from state-of-the-art image search engines by adopting the idea of affinity propagation and design a fast sparse affinity propagation algorithm to find exemplars that best represent the imagesearch results.
- Dense Interaction Learning for Video-based Person Re-identification77
A hybrid framework that takes the principal advantages of both CNN-based and Attention-based architectures to tackle video-based person re-ID difficulties and introduces Spatio-TEmporal Positional Embedding into the DI decoder to investigate the positional relation among the spatial-temporal inputs.
- Deep Saliency Hashing for Fine-Grained Retrieval77
This work proposes a novel deep hashing model, named deep saliency hashing (DSaH), which automatically mines salient regions and learns semantic-preserving hashing codes simultaneously, and performs the best for the fine-grained retrieval task and beats the strongest competitor by approximately 10% on both Stanford Dogs and CUB Birds.
- 77
- 77
- Correlative Linear Neighborhood Propagation for Video Annotation77
This paper proposes a novel method named correlative linear neighborhood propagation to improve annotation performance and demonstrates its effectiveness and efficiency on the Text REtrieval Conference VIDeo retrieval evaluation data set.
- Multi-modality web video categorization77
This paper proposes two novel modalities including a semantic modality and a surrounding text modality, as effective complements to most commonly used low-level features of web video, and demonstrates that the proposed multimodal feature representation is effective for web video categorization.
- 76
- Trinary-Projection Trees for Approximate Nearest Neighbor Search75
This work presents a coordinate-wise enumeration algorithm to find the principal trinary-projection direction and provides an extension using multiple randomized trees for improved performance on large-scale local patch indexing and similar image search.
- Box2Mask: Box-Supervised Instance Segmentation via Level-Set Evolution74
This paper presents a novel single-shot instance segmentation approach, namely Box2Mask, which integrates the classical level-set evolution model into deep neural network learning to achieve accurate mask prediction with only bounding box supervision.
- Transporting Causal Mechanisms for Unsupervised Domain Adaptation73
Transporting Causal Mechanisms (TCM) is proposed, to identify the confounder stratum and representations by using the domain-invariant disentangled causal mechanisms, which are discovered in an unsupervised fashion.
- 73
- Combining global, regional and contextual features for automatic image annotation73
A novel approach to automatic image annotation which combines global, regional, and contextual features by an extended cross-media relevance model that describes the global features as a distribution vector of visual topics and model the textual context as a multinomial distribution.
- AVE73
This paper developed an approach for extracting temporal structure and determining the importance of a video segment in order to facilitate the selection of highlight segments, and extracts temporal structure, beats and tempos from the incidental music.
- 72
- 72
- Dynamic Hypergraph Structure Learning for Traffic Flow Forecasting71
This paper proposes a novel model named Dynamic Hypergraph Structure Learning (DyHSL) for traffic flow prediction, which extracts hypergraph structural information to model dynamics in the traffic networks, and updates each node representation by aggregating messages from its associated hyperedges.
- Video collage: presenting a video sequence using a single image71
This paper presents an automatic procedure for constructing a compact synthesized collage from a video sequence, and shows its effectiveness to present a video in a very compact and visually appealing form while preserving the necessary information to understand the video.
- 70
- Automatic music video generation based on temporal pattern analysis70
A system that automatically generates MV-like videos from personal home videos based on observations that generally there are obvious repetitive visual and aural patterns in MVs is presented.
- CoCo: A Coupled Contrastive Framework for Unsupervised Domain Adaptive Graph Classification68
This paper proposes Coupled Contrastive Graph Representation Learning (CoCo), which extracts the topological information from coupled learning branches and reduces the domain discrepancy with coupled contrastive learning.
- Traffic Flow Prediction with Vehicle Trajectories68
A spatiotemporal deep learning framework that mines the underlying causality of flows from historical vehicle trajectories and incorporates that into road traffic prediction and achieves over 5% error reduction when compared with the state-of-the-art approaches.
- 68
- Contextual in-image advertising67
An innovative contextual advertising system driven by images is proposed, which automatically associates relevant ads with an image rather than the entire text in a Web page and seamlessly inserts the ads in the nonintrusive areas within each individual image.
- 66
- 66
- Modeling social strength in social media community via kernel-based learning65
A kernel-based learning to rank framework for inferring the social strength of Flickr users, which involves two learning stages and is able to conduct collaborative recommendation and collective classification.
- 64
- Offline-Online Associated Camera-Aware Proxies for Unsupervised Person Re-Identification62
This work proposes to split each single cluster into multiple proxies according to camera views, and designs two proxy-level contrastive learning losses that enable the combination of two losses to train a desirable Re-ID model.
- Multiphysical graph neural network (MP-GNN) for COVID-19 drug design62
The proposed multiphysical graph neural network (MP-GNN) model can outperform all existing models as far as the authors know, and is used in coronavirus disease 2019 drug design and demonstrates the great potential of the model for the screening of potential drugs for SARS-CoV-2.
- 62
- Robust Distance Metric Learning with Auxiliary Knowledge ∗62
Two algorithms are derived on the basis of manifold regularization and log-determinant divergence regularization technique, respectively, which can simultaneously exploit label information, unlabeled examples, and the metrics derived from auxiliary data sets and are efficient and flexible.
- 60
- 59
- Semi-automatic video annotation based on active learning with multiple complementary predictors59
It is proved that the samples selected by the proposed scheme are more representative than general active learning scheme, as well as the incremental model adaptation scheme is effective especially when the newly added data size is small.
- Large-scale vehicle trajectory reconstruction with camera sensing network58
VeTrac, a comprehensive system that employs widely deployed traffic cameras as a sensing network to trace vehicle movements and reconstruct their trajectories in a large scale, fuses mobility correlation and vision-based analysis to reduce uncertainties in identifying vehicles.
- Deep Siamese Network with Multi-level Similarity Perception for Person Re-identification58
A novel deep Siamese architecture based on convolutional neural network and multi-level similarity perception for person re-identification achieves splendid performance compared with the current state-of-the-art approaches.
- 58
- The role of attractiveness in web image search58
Experimental results on three large-scale web image search datasets demonstrate that the incorporation of attractiveness can bring more satisfaction to 80% of the users for ranking/reranking search results and 30.5% index coverage improvement for index selection, compared to the conventional relevance based approaches.
- Correlative multilabel video annotation with temporal kernels58
This article proposes another paradigm of the video annotation method that simultaneously annotates the concepts as well as model correlations between them in one step by the proposed Correlative Multilabel (CML) method, which benefits from the compensation of complementary information between different labels.
- Image search by concept map57
A new image search interface is proposed to enable users to formulate a query, called concept map, by intuitively typing textual queries in a blank canvas to indicate the desired spatial positions of the concepts.
- Learning semantic distance from community-tagged media collection57
A semi-definite programming is formulated to encode the above two aspects of criteria to learn the distance metric and it is shown such an optimization problem can be efficiently solved with a closed-form solution.
- 57
- 57
- City brain: practice of large‐scale artificial intelligence in the real world55
The authors introduce current practices of the City Brain platform in a few cities in China, including what they can do to achieve the goal and make it a reality, and focus on the system overview and key technical details of each component of the city Brain system, from cognition to intervention.
- Content-based tag processing for Internet social images55
A comprehensive survey of the technical achievements in the research area of content-based tag processing for social images, covering the research aspects on tag ranking, tag refinement and tag-to-region assignment is provided.
- Cross-Domain Empirical Risk Minimization for Unbiased Long-Tailed Classification54
Cross-Domain Empirical Risk Minimization (xERM) is proposed for training an unbiased test-agnostic model to achieve strong performances on both test distributions, which empirically demonstrates that xERM fundamentally improves the classification by learning better feature representation rather than the "head vs. tail" game.
- CRSSC: Salvage Reusable Samples from Noisy Data for Robust Learning54
This work proposes a certainty-based reusable sample selection and correction approach, termed as CRSSC, for coping with label noise in training deep FG models with web images, which is to additionally identify and correct reusable samples, and then leverage them together with clean examples to update the networks.
- MSRA-USTC-SJTU AT TRECVID 2007: HIGH-LEVEL FEATURE EXTRACTION AND SEARCH54
This paper describes the MSRA-USTC-SJTU experiments for TRECVID 2007 and investigates the benefit of unlabeled data by semi-supervised learning, and the multi-layer (ML) multi-instance (MI) relation embedded in video by MLMI kernel, as well as the correlations between concepts by correlative multi-label learning.
- Million-scale near-duplicate video retrieval system53
A visual word based approach is proposed, which quantizes each video frame into a word and represents the whole video as a bag of words and can respond to a query in 41ms with 78.4% MAP on average.
- 52
- DEAL: An Unsupervised Domain Adaptive Framework for Graph-level Classification52
A novel GNN framework named DEAL is presented by incorporating both source graphs and target graphs, which is featured by two modules, i.e., adversarial perturbation and pseudo-label distilling, which improves the performance of graph classification.
- CPR-GCN: Conditional Partial-Residual Graph Convolutional Network in Automated Anatomical Labeling of Coronary Arteries52
A conditional partial-residual graph convolutional network (CPR-GCN), which takes both position and CT image into consideration, since CT image contains abundant information such as branch size and spanning direction.
- A transductive multi-label learning approach for video concept detection52
This paper presents an effective transductive multi-label classification approach that simultaneously models the labeling consistency between the visually similar videos and the multi- label interdependence for each video.
- 52
- 51
- Sharp Attention Network via Adaptive Sampling for Person Re-Identification51
Extensive experimental evaluations demonstrate the superiority of this new sharp attention model for person re-ID over other related existing, published state-of-the-art works on three challenging benchmarks, including CUHK03, Market-1501, and DukeMTMC-reID.
- Building a comprehensive ontology to refine video concept detection51
This paper presents a comprehensive representation scheme for video semantic ontology in which all the three components are well studied, and leverage LSCOM to construct the concept lexicon, describe concept property as the weights of different modalities which are obtained manually or by data-driven approach.
- Structure-sensitive manifold ranking for video concept detection51
This paper proposes a novel graph-based learning approach, named Structure-Sensitive Manifold Ranking (SSMR), based on a structure-sensitive similarity measure, which takes local distribution differences into account to more accurately measure pairwise similarity.
- DisenSemi: Semi-Supervised Graph Classification via Disentangled Representation Learning50
A novel framework termed DisenSemi is introduced, which learns disentangled representation for semi-supervised graph classification and defines an MI-based disentangled consistency regularization between two models and identifies the corresponding rationale that aligns well with the current graph classification task.
- 50
- Manifold-ranking based video concept detection on large database and feature pool50
A novel manifold- ranking based scheme is proposed, which consists of three major components: feature pool construction, pre-filtering, and manifold-ranking, which is extensible and flexible in terms of adding new features into the feature pool, introducing human interactions on selecting features, and defining new concepts.
- Automatic video annotation by semi-supervised learning with kernel density estimation50
This paper proposes a novel semi-supervised learning algorithm, named Semi Supervised Learning by Kernel Density Estimation (SSLKDE), which is based on a non-parametric method, and therefore the "model assumption" is avoided.
- 50
- Random Boxes Are Open-world Object Detectors49
RandBox is proposed, a Fast R-CNN based architecture trained on random proposals at each training iteration, surpassing existing Faster R-CNN and Transformer based OWOD and significantly outperforms the previous state-of-the-art in all metrics.
- Local Convolutional Neural Networks for Person Re-Identification49
For the task of person re-identification, the local convolutional neural networks (Local CNN) can outperform state-of-the-art methods consistently on three large-scale benchmarks, including Market-1501, CUHK03, and DukeMTMC-ReID.
- 49
- 49
- VideoReach49
This paper presents a novel online video recommendation system called VideoReach, which alleviates users' efforts on finding the most relevant videos according to current viewings without a sufficient collection of user profiles as required in traditional recommenders.
- Homography Loss for Monocular 3D Object Detection48
A differentiable loss function, termed as Homography Loss, is proposed to achieve the goal, which exploits both 2D and 3D information, aiming at balancing the positional relationships between different objects by global constraints, so as to obtain more ac-curately predicted 3D boxes.
- Attention-guided Temporally Coherent Video Object Matting48
A novel deep learning-based video object matting method that can achieve temporally coherent matting results by designing a novel loss term to train the attention weights, which drastically boosts the video matting performance.
- 48
- Category Dictionary Guided Unsupervised Domain Adaptation for Object Detection48
A category dictionary guided (CDG) UDA model for cross-domain object detection is proposed, which learns category-specific dictionaries from the source domain to represent the candidate boxes in target domain and a residual weighted self-training paradigm is developed to implicitly align source and target domains for detection model training.
- 47
- Video eCommerce47
An advertising system named Video eCommerce to exhibit appropriate product ads to particular users at proper time stamps of videos, which takes into account video semantics, user shopping preference and viewing behavior feedback by a two-level strategy is proposed.
- 47
- 47
- TGNN: A Joint Semi-supervised Framework for Graph-level Classification46
This paper proposes a novel semi-supervised graph classification framework called Twin Graph Neural Network (TGNN), which has two twin modules that collaborate with each other by exchanging instance similarity knowledge to fully explore the structure information of both labeled and unlabeled data.
- Self-Regulation for Semantic Segmentation46
This paper seeks reasons for the two major failure cases in Semantic Segmentation (SS): 1) missing small objects or minor object parts, and 2) mislabeling minor parts of large objects as wrong classes and introduces several Self-Regulation (SR) losses for training SS neural networks.
- 45
- 45
- 45
- Online multi-label active annotation45
This paper proposes a scalable framework for annotation-based video search, as well as a novel approach to enable large-scale semantic concept annotation, that is, online multi-label active learning, scalable to both the video sample dimension and concept label dimension.
- 45
- CIMON: Towards High-quality Hash Codes44
A new method named Comprehensive sImilarity Mining and cOnsistency learNing (CIMON) is proposed, which uses global refinement and similarity statistical distribution to obtain reliable and smooth guidance and both semantic and contrastive consistency learning are introduced.
- Apparel-Invariant Feature Learning for Person Re-Identification44
The semi-supervised Apparel-invariant Feature Learning (AIFL) framework is proposed to learn an apparel- Invariant pedestrian representation using images of the same person wearing different clothes, and an unsupervised apparel-simulation GAN (AS-GAN) to synthesize cloth-changing images according to the target cloth embedding.
- A Domain Robust Approach For Image Dataset Construction44
This work proposes a novel image dataset construction framework which can generalize well to unseen target domains and formulate image filtering as a multi-instance learning (MIL) problem with constrained positive bags.
- 44
- 44
- 43
- Towards Counterfactual Image Manipulation via CLIP42
A novel contrastive loss is designed that exploits predefined CLIP-space directions to guide the editing toward desired directions from different perspectives and achieves accurate and realistic editing while driving by target texts with various counterfactual concepts.
- Towards Fine-Grained Human Pose Transfer With Detail Replenishing Network42
This work develops a more challenging yet practical HPT setting, termed as Fine-grained Human Pose Transfer (FHPT), with a higher focus on semantic fidelity and detail replenishment, and builds up a complete suite of fine- grained evaluation protocols to address the challenges of FHPT in a comprehensive manner.
- 42
- Social Image Search with Diverse Relevance Ranking42
A diverse relevance ranking scheme which is able to simultaneously take relevance and diversity into account and takes advantage of both the content of images and their associated tags.
- Typicality ranking via semi-supervised multiple-instance learning42
A novel typicality ranking scheme for categorizing natural scenes through a two-stage semi-supervised multiple-instance learning method that only needs image-level labels instead of region-level ones in the training stage and is fully automated and no human feedback is required.
- Near-lossless semantic video summarization and its applications to video analysis41
This article presents a novel approach, called Near-Lossless Semantic Summarization (NLSS), to summarize a video stream with the least high-level semantic information loss by using an extremely small piece of metadata.
- 40
- 40
- Revisiting Knowledge Distillation: An Inheritance and Exploration Framework39
This paper proposes a novel inheritance and exploration knowledge distillation framework (IE-KD), in which a student model is split into two parts - inheritance and Exploration, which can be a general technique to improve the student network to achieve SOTA performance.
- 38
- 38
- 38
- To learn representativeness of video frames38
This paper developed a method to examine and evaluate the representativeness of video frames based on learning users' perceptive evaluations, and observed that users have similar tendency in selecting the most representative frame for a certain video segment.
- SSAH: Semi-Supervised Adversarial Deep Hashing with Self-Paced Hard Sample Generation37
A novel Semi-supervised Self-pace Adversarial Hashing method, named SSAH, to solve the above problems in a unified framework and can significantly improve state-of-the-art models on both the widely-used hashing datasets and fine-grained datasets.
- ImageSense37
A contextual advertising system driven by images is presented, which automatically associates relevant ads with an image rather than the entire text in a Web page and seamlessly inserts the ads in the nonintrusive areas within each individual image.
- Visual tag dictionary: interpreting tags with visual words37
A corpus named visual tag dictionary is introduced, which interprets each tag with visual words and is constructed in a fully automatic way by exploring the tagged image data on the Internet.
- Learning With Imbalanced Noisy Data by Preventing Bias in Sample Selection36
This work proposes Class-Balance-based sample Selection (CBS) to prevent the tail class samples from being neglected during training, and introduces the Average Confidence Margin (ACM) metric to measure the quality of corrected labels by leveraging the model's evolving training dynamics.
- Concentrated Local Part Discovery With Fine-Grained Part Representation for Person Re-Identification36
- 36
- Bridging low-level features and high-level semantics via fMRI brain imaging for video classification36
There are meaningful couplings between brain's fMRI responses and video stimuli, suggesting the validity of linking semantics and low-level features via fMRI; and the computationally learned low- level feature sets from fMRI-derived semantic features can significantly improve the classification of video categories in comparison with that based on original low-levels features.
- Retagging social images based on visual and semantic consistency36
This paper proposes a social image ``retagging'' scheme that aims at assigning images with better content descriptors that is formulated as an optimization framework based on the consistency between ``visual similarity'' and ``semantic similarity'' in social images.
- 36
- 36
- A Statistical Approach to Mining Semantic Similarity for Deep Unsupervised Hashing35
A general distribution-based metric is proposed to depict the pairwise distance between images and is applied to deep unsupervised hashing, resulting in Distribution-based similArity sTructure rEconstruction (DATE), which can generate more accurate semantic similarity information by using non-parametric ball divergence.
- 35
- Image tag refinement by regularized latent Dirichlet allocation34
This paper proposes to do tag refinement from the angle of topic modeling and presents a novel graphical model, regularized latent Dirichlet allocation (rLDA), where tag similarity and tag relevance are jointly estimated in an iterative manner, so that they can benefit from each other.
- 34
- Automatically converting photographic series into video34
A novel way to browse a series of otogras is proposed, which can be regarded as a system exploring the new medium between otogra and video.
- 33
- Video Histogram: A Novel Video Signature for Efficient Web Video Duplicate Detection33
This paper proposes a novel signature-based approach for duplicate video comparison that requires less storage and computation cost than previous methods and is both efficient and effective for web video duplicate detection.
- 32
- 32
- Annotating personal albums via web mining32
A multi-graph similarity propagation based semi-supervised learning (MGSP-SSL) algorithm is proposed to suppress the noises in the initial annotations from the Web, and experiments show that the proposed approach outperforms existing annotation methods.
- An Attention-Based Decision Fusion Scheme for Multimedia Information Retrieval32
A novel decision fusion scheme based on the psychological observations on human beings' visual and aural attention characteristics, which combines a set of decisions obtained from different data sources or features to generate better decision result is proposed.
- 31
- 31
- 31
- 30
- 30
- 29
- Landmarks Detection with Anatomical Constraints for Total Hip Arthroplasty Preoperative Measurements29
An automatic measurement method by detecting anatomical landmarks with the latest deep learning approaches for total hip arthroplasty by imposing the relative position constraints on each landmark by defining edges among landmarks according to the clinical significance.
- 29
- Large-scale robust visual codebook construction29
An improved approximate K-means is proposed, by leveraging the assignment information in the history, namely the previous iterations, to improve the assignment precision by further randomizing the employed approximate nearest neighbor search in each iteration.
- VideoSense29
VideoSense is able to detect a set of candidate ad insertion points based on content analysis, select a list of relevant candidate ads ranked according to textual relevance, and find the best match between insertion points and ads which maximizes the overall multimodal relevance.
- 28
- 28
- Automated Segmentation Of Pulmonary Lobes Using Coordination-Guided Deep Neural Networks28
This work introduces coordination-guided convolutional layers (CoordConvs) that generate additional feature maps of the positional information of pulmonary lobes that reduce the misclassification of different lobes.
- Melog: mobile experience sharing through automatic multimedia blogging28
This paper will propose a "mobile + cloud" system enabling rapid and near-realtime experience sharing through automatic blogging and micro-blogging, which are based on multi-modal media content analyses and syntheses.
- 27
- TSNAdb v2.0: The Updated Version of Tumor-Specific Neoantigen Database27
An updated version of the Tumor-Specific NeoAntigen Database (TSNAdb v2.0) is provided with several new features including taking stricter criteria for neoantigen identification and providing predicted neoantigens derived from three types of somatic mutations.
- AdOn: toward contextual overlay in-video advertising27
AdOn represents one of the first attempts towards contextual overlay video advertising by leveraging information retrieval and multimedia content analysis techniques and is superior to existing advertising approaches in terms of ad relevance and user experience.
- Molecular persistent spectral image (Mol-PSI) representation for machine learning models in drug design26
An equal-sized molecular 2D image representation, known as the molecular persistent spectral image (Mol-PSI), is proposed, and combined with CNN model for AI-based drug design and molecular data analysis, which are better than all traditional machine learning models.
- 26
- Interactive browsing via diversified visual summarization for image search results26
A novel search result summarization approach is introduced and an interactive browsing scheme, based on a tree structure for organizing the images obtained from the summarization approaches, is proposed to enable users to intuitively and conveniently browse the image search results.
- Transductive multi-label learning for video concept detection26
This paper proposes an effective transductive multi-label classification approach, which simultaneously models the labeling consistency between the visually similar videos and the multi- labels interdependence for each video in an integrated framework and compares the performance between the proposed approach and several representative transductives single-label and supervised multi- label classification approaches.
- 26
- 26
- An online-optimized incremental learning framework for video semantic classification26
A new online semantic classification framework, termed OOIL (for Online-Optimized Incremental Learning), is proposed, in which two sets of optimized classification models are online trained by sufficiently exploiting both local and global statistic characteristics of videos.
- Semi-Supervised Semantic Segmentation With Multi-Constraint Consistency Learning25
A feature knowledge alignment (FKA) strategy to promote the feature consistency learning of the encoder from image-augmentation and a self-adaptive intervention (SAI) module to increase the discrepancy of aligned intermediate feature representations, promoting Feature-perturbation based Prediction consistency learning.
- Towards Precise Intra-camera Supervised Person Re-Identification25
This paper proposes jointly learned camera-specific non-parametric classifiers, together with a hybrid mining quintuplet loss, to perform intra-camera learning and shows that this approach outperforms all existing ICS works by a great margin.
- 25
- Bridging the Web Data and Fine-Grained Visual Recognition via Alleviating Label Noise and Domain Mismatch24
This work mainly focuses on two critical issues including "label noise" and "domain mismatch" in the web images, and proposes an end-to-end deep denoising network (DDN) model to jointly solve these problems in the process of web images selection.
- A new web-supervised method for image dataset constructions24
A novel automatic image dataset construction framework is proposed by employing multiple query expansions, which indicates the effectiveness of the method is superior to weak supervised and web supervised state-of-the-art methods.
- Optimizing video search reranking via minimum incremental information loss24
It is proposed to directly optimize video search reranking from a novel viewpoint of information theory, that is, to identify an optimal set of correctly-ranked document pairs which maximally preserves the relevant information and simultaneously carries the irrelevant information as little as possible.
- 24
- Structure and event mining in sports video with efficient mosaic24
A mosaic based approach to key-event as well as structure mining, which is regarded as a complementary view for sports video analysis and the detection of key-plays and key-events in the specific-domain of soccer videos have proved its effectiveness.
- 23
- Layout Style Modeling for Automating Banner Design23
A novel style transfer technique is invented so that creating a multi-size style becomes as easy as designing a single banner, and a robust and easy-to-use layout style model is built, upon which the banner design is automated.
- Prajna: Towards Recognizing Whatever You Want from Images without Image Labeling23
The intrinsic characteristics of Internet image data are analyzed, and ways to mine accurate and informative infor-mation from those data to build a training set, which is then used to train image recognition models.
- 23
- Hybrid image summarization23
A novel approach, called Homogeneous and Heterogeneous Message Propagation (H2MP), which extends affinity propagation that only works over homogeneous relations to heterogeneous relations and the summary obtained by this approach is both visually and semantically satisfactory.
- Video annotation based on temporally consistent Gaussian random field23
A novel method for automatically annotating video semantics, called temporally consistent Gaussian random field (TCGRF), which adapts the temporal consistency property of video data into graph-based semi-supervised learning to improve the annotation results.
- 23
- 22
- Improved Deep Unsupervised Hashing via Prototypical Learning22
This research introduces both feature prototypes and hashing prototypes to model the underlying semantic structures of the images in both deep feature space and hash code space, and imposes a smoothness constraint to regularize the consistency of the global structures in two spaces through semantic prototypical consistency learning.
- On Non-Random Missing Labels in Semi-Supervised Learning22
This work integrates CAP and CAI into a Class-Aware Doubly Robust (CADR) estimator for training an unbiased SSL model and significantly outperforms existing baselines but also surpasses other label bias removal SSL methods.
- Effective music tagging through advanced statistical modeling22
A novel system called MMTagger is introduced that effectively integrates both multimodal and temporal information in the representation of music signal and seamlessly combines Multiple Gaussian Mixture Models (GMMs) and Support Vector Machine (SVM) into a single framework.
- Interactive image search by 2D semantic map22
A novel interactive image search system, image search by 2D semantic map, that enables users to indicate what semantic concepts are expected to appear and even how these concepts are spatially distributed in the desired images is presented.
- Near-lossless video summarization22
A new system for video summarization, called "Near-Lossless Video Summarization" (NLVS), which is able to summarize a video stream with the least information loss by using an extremely small piece of metadata.
- AdOn22
AdOn represents one of the first attempts towards intelligent overlay video advertising by leveraging video content analysis techniques and is able to automatically detect a set of spatio-temporal nonintrusive positions and associate the contextually relevant ads with these positions.
- LazyCut22
LazyCut provides a semi-automatic video authoring and sharing system that significantly reduces users' efforts in video editing while preserving sufficient flexibility and personalization.
- SPORT: A Subgraph Perspective on Graph Classification with Label Noise21
This work investigates the problem of graph classification with label noise, which is demanding because of the complex graph representation learning issue and serious memorization of noisy samples, and presents a novel approach called Subgraph Set Network with Sample Selection and Consistency Learning (SPORT) for this problem.
- 21
- 21
- 21
- 21
- 20
- PyRetri: A PyTorch-based Library for Unsupervised Image Retrieval by Deep Convolutional Neural Networks20
PyRetri is an open source library for deep learning based unsupervised image retrieval that encapsulates the retrieval process in several stages and provides functionality that covers various prominent methods for each stage.
- Discriminative Coronary Artery Tracking via 3D CNN in Cardiac CT Angiography20
Experimental results on a private collected clinical dataset demonstrate the effectiveness of the proposed discriminative coronary artery tracking method (DCAT), which outperforms state-of-the-art methods on the public dataset in CAT08 challenge.
- 20
- Scalable clip-based near-duplicate video detection with ordinal measure20
This paper proposes to use Multi-Probe Locality Sensitive Hashing (MPLSH) to index the video clips for fast similarity search and high recall and is able to filter out a large number of dissimilar clips from video database.
- Token Embeddings Alignment for Cross-Modal Retrieval19
A novel Token Embeddings AlignMent (TEAM) block is proposed, it first explicitly aligns visual tokens and textual tokens, then produces token-level matching scores to measure fine-grained similarity between input image and text.
- SP-ViT: Learning 2D Spatial Priors for Vision Transformers19
This work presents Spatial Prior-enhanced Self-attention (SP-SA), a novel variant of vanilla Self-Attention (SA) tailored for vision transformers and proposes the SP-ViT family, which consistently outperforms other ViT models with similar GFlops or parameters.
- 19
- Extracting Privileged Information from Untagged Corpora for Classifier Learning19
A new instance-level multi-instance learning (MIL) model is proposed to enhance classifier learning by extracting PI from untagged corpora, which can effectively eliminate the dependency on manually labeled data.
- Multimedia tagging19
This tutorial aims to provide a comprehensive coverage on the evolution of research for developing multimedia tagging technologies and identify a range of major challenges for the further scholarly study in the coming years.
- MSRA atT TRECVID 2008: High-Level Feature Extraction and Automatic Search.19
This paper describes the MSRA experiments for TRECVID 2008, which representatively investigated the benefit of global and local low-level features by a variety of learning-based methods, including supervised and semi-supervised learning algorithms.
- Video collage19
Video Collage system is presented, which automatically constructs a compact and visually appealing synthesized collage from a video sequence for efficient video browsing and provides a novel user interface that enables users to browse video content in a variety of more efficient ways in contrast to many existing approaches to video browsing.
- 19
- DANCE: Learning A Domain Adaptive Framework for Deep Hashing18
A novel method named unbiaseD duAl hashiNg Contrastive lEarning (DANCE) for domain adaptive image retrieval to perform contrastive learning on hash codes from both instance level and prototype level and enhances domain alignment in the Hamming space while maximizing the model capacity.
- Deep Adaptive Graph Clustering via von Mises-Fisher Distributions18
Deep Adaptive Graph Clustering via von Mises-Fisher distributions is proposed, namely DAGC, which can simultaneously improve the intra-cluster compactness and inter-cluster heterogeneity.
- 18
- 18
- Accessible image search18
This paper introduces a scheme named Accessible Image Search (AIS), which takes into account the colorblind accessibilities of the returned results, and proposes an efficient recoloring algorithm to modify the colors of the images such that they can be better perceived by colorblind users.
- Video semantic analysis based on structure-sensitive anisotropic manifold ranking18
A novel graph-based SSL framework, named structure-sensitive anisotropic manifold ranking (SSAniMR), is deduced from PDE-based anisotrop diffusion, which significantly outperforms existing graph- based SSL methods and is effective for video semantic annotation.
- Free-Shaped Video Collage18
This paper forms the generation of FS-Collage as an energy minimization problem and solves the problem by designing a random sampling process and shows that the experiment results show that the FS- collage achieves satisfying performance.
- Refining video annotation by exploiting pairwise concurrent relation18
This paper proposes to refine video annotation by leveraging the pairwise concurrent relation among video concepts, explicitly modeled by a concurrent matrix and then a propagation strategy is adopted to refine the annotations.
- Invariant Training 2D-3D Joint Hard Samples for Few-Shot Point Cloud Recognition17
This work tackles the data scarcity challenge in few-shot point cloud recognition of 3D objects by using a joint prediction from a conventional 3D model and a well-trained 2D model, and proposes an invariant training strategy, called INVJOINT, which can learn more collaborative 2D and 3D representations for better ensemble.
- 17
- Previewer for Multi-Scale Object Detector17
A novel light-weight previewer block is proposed, which previews the objectness probability for the potential regression region of each prior box, using the stronger features with larger receptive fields and more contextual information for better predictions.
- DeepProduct17
Extensive experiments on challenging real-world datasets demonstrate that both the robust contrastive loss and the multitask fine-tuning scheme are effective, leading to very promising results with a time cost suitable for mobile product search scenarios.
- 17
- 17
- 16
- 16
- Meta Clustering Learning for Large-scale Unsupervised Person Re-identification16
This paper makes the first attempt to the large-scale U-ReID and proposes a "small data for big task" paradigm dubbed Meta Clustering Learning (MCL), which only pseudo-labels a subset of the entire unlabeled data via clustering to save computing for the first-phase training.
- The City Brain: Towards Real-Time Search for the Real-World16
This talk will introduce current practices of the City Brain platform in a few cities in China, including what can be done to achieve the goal and make it a reality, and focus on visual search technologies and applications that can apply on the city data.
- Contextual image retrieval model16
A contextual image retrieval model based on the language modeling approach that considers the bounding box as an uncertain observation of the latent search intention and the saliency map detected for the query image as a prior and a search intention score inferred per visual word and used to weight the ROI and the context for a better estimation of the query language model.
- 16
- 16
- 15
- Pairwise VLAD Interaction Network for Video Question Answering15
A pairwise VLAD interaction mechanism is proposed to better exploit complementary information for each pair of modality descriptors, avoiding modeling uninformative individual relations, and exploring both inter- and intra-modality relations simultaneously.
- Dual Graph for Traffic Forecasting15
A novel dual graph framework, called DualGraph, is proposed, to model the propagation behavior of traffic on road networks, and shows that even for node or edge traffic forecasting alone, the model still outperforms compared ones, especially for long term prediction.
- Image retrieval with query-adaptive hashing15
Extensive experiments over three benchmark image datasets well demonstrate the superiority of the proposed query-adaptive hashing method over the state-of-the-art ones in terms of retrieval accuracy.
- Contextual image search15
The key idea is to use the search-to-annotation technique and the contextual textual query mining scheme to determine the corresponding textual query, to finally get semantically similar search results.
- Metric learning with feature decomposition for image categorization15
An approach named distance metric learning with feature decomposition (DMLFD) that can reduce the computational costs ofdistance metric learning methods as well as improve their performance for image categorization is proposed.
- Visual Reranking with Local Learning Consistency15
This paper proposes to use local learning regularizer to model the multiple-wise consistency, by formulating the consistent score estimation over a local area into a learning problem.
- Collaborative learning for image and video annotation15
This paper assumes that visually similar concepts should share resembled detectors, and proposes Multi-Granularity Boosting strategy, where each granularity corresponds to a specific balance between collaboration and discrimination for Collaborative Learning.
- INTERACTIVE VIDEO ANNOTATION BY MULTI-CONCEPT MULTI-MODALITY ACTIVE LEARNING15
A multi-concept multi-modality active learning method for video annotation in which multiple concepts and multiple modalities can be simultaneously taken into consideration is proposed.
- Interactive video authoring and sharing based on two-layer templates15
LazyMedia provides a semiautomatic video authoring and sharing system that significantly reduces users' efforts in video editing while preserving sufficient flexibility and personalization.
- Personal media sharing and authoring on the web15
Three primary technologies enable this end-to-end system, including scalable video coding, intelligent multimedia content analysis, and template-based media authoring, which reduces the workload of media sharing and authoring.
- 15
- 14
- 14
- 14
- Query sampling for ranking learning in web search14
A greedy algorithm to sample the queries, by simultaneously taking the query density, difficulty and diversity into consideration, shows that the proposed method can lead to a more informative training set for building an effective model.
- 14
- MILC2: A Multi-Layer Multi-Instance Learning Approach to Video Concept Detection14
A novel algorithm named MILC2 (Multi-Layer Multi-Instance Learning with Inter-layer Consistency Constraint.) is proposed to tackle the ambiguity propagation problem, in which an inter-layer consistency constraint is explicitly introduced to measure the disagreement of inter-layers, and thus the MI relation is better modeled.
- Multi-layer multi-instance kernel for video concept detection14
This paper forms video concept detection as an MLMI learning problem in which a rooted tree with MLMI nature embedded is devised to represent a video segment and constructs a novel MLMI kernel to measure the similarities between the instances in the same and different layers.
- Spatio-temporal quality assessment for home videos14
This paper presents a novel spatio-temporal quality assessment scheme in terms of low-level content features for home videos, aiming at achieving the best quality while simultaneously preserving the most informative content.
- P-Karaoke14
A personalized Karaoke system, <i>P-Karaoke</i>, is proposed, which uses personal home videos and photographs, which are automatically selected from users' multimedia database according to their content, users' preferences or music.
- Criterion-based Heterogeneous Collaborative Filtering for Multi-behavior Implicit Recommendation13
A novel non-sampling learning framework named Criterion-guided Heterogeneous Collaborative Filtering (CHCF), which integrates criterion learning and user preference learning into a unified framework, which can be trained jointly for the interaction prediction of the target behavior.
- 13
- Concept Detection based on Multi-label Classification and Image Captioning Approach - DAMO at ImageCLEF 2019.13
This paper illustrates the proposed method to address the issue, based on multi-label classification model and CNN-LSTM architecture with attention mechanism, and performs a detailed analysis and processing for the overall dataset.
- Homocentric Hypersphere Feature Embedding for Person Re-Identification13
A novel homocentric hypersphere embedding scheme is proposed to decouple magnitude and orientation information for both feature and weight vectors, and reformulate the triplet loss and the softmax loss to their angular versions and combine them into an angular discriminative loss.
- Learning Feature Embedding with Strong Neural Activations for Fine-Grained Retrieval13
An effective feature embedding is proposed by simultaneously encoding original global features and discriminative local features, in which the local features are extracted by exploiting strong neural activations on the last convolutional layer, which is the key factor to improve retrieval precision.
- 13
- Intelligent photo clustering with user interaction and distance metric learning13
Experimental results on multiple photo albums demonstrated that this approach is able to improve automatic photo clustering results, and by exploring distance metric learning, this method is much more effective than pure manual adjustments ofphoto clustering.
- Accessible image search for colorblindness13
An intelligent system that accommodates colorblind users in image search that takes into account the colorblind accessibilities of the returned results and proposes an efficient recoloring algorithm to modify the colors of the images such that they can be better perceived by color blind users.
- Multiple-Instance Active Learning for Image Categorization13
This paper proposes a set kernel based classifier, based on which, a unified bag and/or instance selection criterion and an integrated learning algorithm are built and shows that selecting both bags and instances outperforms the other two strategies.
- 13
- 13
- 13
- Why We Feel: Breaking Boundaries in Emotional Reasoning with Multimodal Large Language Models12
This work presents EIBench, a large-scale benchmark encompassing 1615 basic EI samples and 50 complex EI samples featuring multifaceted emotions, and proposes a Coarse-to-Fine Self-Ask (CFSA) annotation pipeline, which guides Vision-Language Models (VLLMs) through iterative question-answer rounds to yield high-quality labels at scale.
- 12
- HARR: Learning Discriminative and High-Quality Hash Codes for Image Retrieval12
A novel unsupervised hashing method called HARR is proposed to learn discriminative and high-quality hash codes and comprehensively explore semantic similarity structure, which adopts the Winner-Take-All hash to model the similarity structure.
- Stylized Adversarial AutoEncoder for Image Generation12
An autoencoder-based generative adversarial network (GAN) for automatic image generation, which is called "stylized adversarial autoenCoder", which achieves superior results for image generation as well as remarkably improves the corresponding supervised recognition task.
- Social Friend Recommendation Based on Network Correlation and Feature Co-Clustering12
This paper proposes a two-stage procedure for more accurate friend recommendation based on the relationship of different social networks, and co-clustering is applied to the tag and image information of the list to refine the recommendation result in the first stage.
- Community Discovery from Social Media by Low-Rank Matrix Recovery12
This paper presents a novel approach to discovering communities from social media, including the group membership and user friend structure, by exploring a low-rank matrix recovery technique for Flickr, and proposes an iterative algorithm to recover the true community indicator matrix.
- Efficient semantic annotation method for indexing large personal video database12
A novel annotation framework based on active learning and semi-supervised ensemble method, which is specially designed for personal video database, shows that the proposed method performs superior to both existing semi- supervised learning algorithms and the general active learning algorithms in terms of annotation accuracy and performance stability.
- Microsoft Research Asia TRECVID 2006 High-Level Feature Extraction and Rushes Exploitation.12
The MSRA experiments for TRECVID 2006 were described, including details of the approaches and performance analyses for high-level feature extraction task and rushes exploitation task, and the approaches to concept modeling were validated.
- Tracking users' capture intention12
A comprehensive and extensible scheme consisting of video structuring, intention oriented feature analysis, as well as intention unit segmentation and classification is proposed to mine the users' capture intention.
- Tracking concept drifting with an online-optimized incremental learning framework12
Experiments on the data set of TREC Video Retrieval Evaluation 2004 not only demonstrate the inside concept drifting process of the learning system, but also prove that the proposed learning framework is promising for tackling the issue of concept drifting.
- HEART: Towards Effective Hash Codes under Label Noise11
This paper focuses on the problem of learning to hash with label noise and proposes a novel method called HEART, a holistic framework which explores latent semantic distributions to select both clean samples and pairs of high confidence for mitigating the impacts of label noise.
- Using generative adversarial networks for genome variant calling from low depth ONT sequencing data11
This work trained a generative adversarial network for low depth variant calling that can project high depth sequencing information from low depth data and validated the trained LDV-Caller model on 157 public Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) samples.
- Graph-Induced Contrastive Learning for Intra-Camera Supervised Person Re-Identification11
This article first formulate the cross-camera ID association task as a graph partitioning problem subjected to ICS-specific constraints and design a greedy agglomeration algorithm to solve it, and proposes a graph-induced contrastive loss that unifies both intra- and inter-camera learning into a contrastive learning framework to learn a Re-ID model.
- Asynchronous Teacher Guided Bit-wise Hard Mining for Online Hashing11
A novel online hashing method to handle the catastrophic forgetting problem, and a novel semantic loss (termed bit-wise attention loss) to dynamically focus on hard samples of each bit during training.
- A Multi-Task Learning Framework for Extracting Bacteria Biotope Information11
This paper presents a novel transfer multi-task learning method for Bacteria Biotope rel+ner task at BioNLP-OST 2019 and achieves the best performance on all metrics (including slot error rate, precision and recall).
- Progressive Transfer Learning for Person Re-identification11
This paper introduces a novel network structure called Batch-related Convolutional Cell (BConv-Cell), which progressively collects the global information of the dataset into a latent state and uses this latent state to rectify the extracted feature.
- Multi-level Similarity Perception Network for Person Re-identification11
A novel deep Siamese architecture based on a convolutional neural network and multi-level similarity perception for the person re-identification (re-ID) problem and final feature embedding by simultaneously encoding original global features and discriminative local features is proposed.
- A unified context model for web image retrieval11
This article evaluated the proposed approach on a representative product image dataset collected from the web and demonstrated that the inclusion of the local and global query contexts significantly improves the performance of QBE-based web image retrieval.
- Tag-Based Social Image Search: Toward Relevant and Diverse Results11
This chapter presents a diverse relevance ranking scheme which simultaneously takes relevance and diversity into account by exploring the content of images and their associated tags.
- Visual quality assessment for web videos11
A first attempt towards assessing web videos in terms of visual quality with significant tests on 30k web videos is reported, showing some preliminary results for application to filtering and re-ranking of retrieved web videos.
- Multi-video synopsis for video representation11
This paper describes a novel approach to present the dynamics of multiple videos simultaneously, aiming at a less intrusive viewing experience, using an efficient Viterbi beam search algorithm and an informative blending algorithm that naturalizes the connecting boundary between different videos.
- 11
- 11
- To construct optimal training set for video annotation11
This paper proposes a set of optimization rules to capture the most distribution information of the whole data for a training set with a given size, and demonstrates that these rules are effective for training set construction for video annotation, and significantly outperform random training set selection.
- An Evaluation of Large Language Models in Bioinformatics Research10
This work studies the performance of large language models like GPT variants on a wide spectrum of crucial bioinformatics tasks, indicating that, given appropriate prompts, LLMs like GPT variants can successfully handle most of these tasks.
- 10
- 10
- Color filter for image search10
A color filter scheme for image search is shown, which enables users to find images containing objects or scenes with their interested color, one of the most crucial cues in describing visual contents.
- Interactive mobile visual search for social activities completion using query image contextual model10
- 10
- FAMER: Making Multi-Instance Learning Better and Faster10
FAMER constructs a Locally Sensitive Hashing based similarity measure for multi-instance framework, and represents each bag as a histogram by embedding instances within the bag into an auxiliary space, which captures the correspondence information between two bags.
- Tagging tags10
An automatic scheme called tagging tags is proposed to supplement semantic image descriptions by associating a group of property tags with each existing tag, which enable much more precise image search especially when certain tag properties are included in the query.
- Smart batch tagging of photo albums10
A smart batch tagging scheme that aims at facilitating users in album tagging, where the number of tagged photos is significantly reduced and a good trade-off between manual efforts and tagging performance can be achieved.
- 10
- 10
- 10
- Semi-automatic video semantic annotation based on active learning10
A representative active learning method is proposed, which local consistency of video content is effectively taken into consideration and the main idea is to exploit the global and local statistical characteristics of videos, and the temporal relationship between shots.
- 9
- A Dual-channel Semi-supervised Learning Framework on Graphs via Knowledge Transfer and Meta-learning9
A dual-channel framework including a supervised model for learning the label probability of nodes and an unsupervised model for extracting information from massive unlabeled graph data is devised, and a knowledge transfer head is proposed to bridge the gap between the generalization and fitting capability of the two models.
- 9
- 9
- FGAGT: Flow-Guided Adaptive Graph Tracking.9
This article proposes the FGAGT tracker, which reaches the level of state-of-the-art, where the MOTA index exceeds FairMOT by 2.5 points, and CenterTrack by 8.4 points on the MOT17 dataset.
- Online Multi-Label Active Learning for Large-Scale Multimedia Annotation9
This paper proposes a scalable framework for annotation-based video search, as well as a novel approach to enable large-scale semantic concept annotation, that is, online multi-label active learning, scalable to both the video sample dimension and concept label dimension.
- 9
- 9
- Automatic Home Video Editing9
An optimization-based system for editing home video in an automatic manner that automatically selects highlight segments from a set of raw home videos and aligns them with a given piece of incidental music to create an edited video segment with a desired length, which is based on the content of the video and the incidental music.
- Video Concept Detection Using Support Vector Machines – TRECVID 2007 Evaluations9
This report describes video concept detection using Support Vector Machine (SVM) over TRECVID 2007 corpus with mainly focus on data preparation for training concept detectors, as well as the preparation of auxiliary training data by using TREC VID 2005 data.
- Towards content-based relevance ranking for video search9
This paper developed an integrated ranking approach using video metadata and content information of shots, which achieves improved ranking performance and introduces machine learning into the ranking system, and compares them with IR-model (information retrieval model) based method.
- 8
- 8
- Pursuing Knowledge Consistency: Supervised Hierarchical Contrastive Learning for Facial Action Unit Recognition8
A supervised hierarchical contrastive learning method (SupHCL) is proposed for AU recognition to pursue knowledge consistency among different facial images and different AUs, which is orthogonal to methods focusing on network architecture design.
- SpineOne: A One-Stage Detection Framework for Degenerative Discs and Vertebrae8
This work proposes a one-stage detection framework termed SpineOne to simultaneously localize and classify degenerative discs and vertebrae from magnetic resonance imaging (MRI) slices and demonstrates that this approach surpasses existing methods by a large margin.
- TapTell: Interactive visual search for mobile task recommendation8
This paper develops TapTell, an exemplary real application on Windows Phone seven, by taking advantages of user interaction and rich context to enable interactive visual searches and contextual recommendations, and compared the results to state-of-the-art algorithms for image retrieval.
- Study on the combination of video concept detectors8
This paper proposes a Bottom-Up Incremental Fusion (BUIF) method which keeps selecting the detectors with lowest performance for combination which has demonstrated the superiority of the proposed incremental fusion approach.
- 8
- 8
- Video booklet: a natural video searching and browsing interface8
A novel system, named Video Booklet, which enables efficient and natural personal video browsing and searching, and builds a seamless bridge between digital videos and analog albums.
- 8
- UCDR-Adapter: Exploring Adaptation of Pre-Trained Vision-Language Models for Universal Cross-Domain Retrieval7
The UCDR-Adapter is proposed, which enhances pre-trained models with adapters and dynamic prompt generation through a two-phase training strategy and dynamically adapts to evolving data distributions, enhancing both flexibility and generalization.
- 7
- 7
- Decouple co-adaptation: Classifier randomization for person re-identification7
This paper proposes a very simple and effective training method, called DeAda, to construct a series of weak classifiers during training by randomization of parameters, so that optimization on the feature extractor could be strengthened in the training stage.
- Extracting Visual Knowledge from the Internet: Making Sense of Image Data7
This work evaluates the proposed webly supervised approach on the benchmark Pascal VOC 2007 dataset and the results demonstrates the superiority of the method over many other state-of-the-art methods in image data collection.
- Introduction to the Special Section on Visual Computing in the Cloud: Fundamentals and Applications7
This Special Section aims to address fundamental and practical aspects of visual computing in the cloud, such as how to build cloud platforms that can cope with seemingly unlimited supply of content coming from traditional media sources as well as new media uploaded to the Internet.
- 7
- Scalable similar image search by joint indices7
This work presents an efficient and effective indexing algorithm, neighborhood graph index, which is suitable for millions of images, and uses it to organize joint inverted indices to search over billions of images.
- 7
- Video-based image retrieval7
A novel QBE-based image retrieval system where users are allowed to submit a short video clip as a query to improve the retrieval reliability and improve the relevance of the retrieved results is proposed.
- Graph-Based Pairwise Learning to Rank for Video Search7
A novel graph-based learning to rank (GLRank) for video search by leveraging the vast amount of unlabeled samples to address the problem of lacking training samples labeled with relevancy degree or ranking orders.
- GameSense: game-like in-image advertising7
A novel game-like advertising system called GameSense, which is driven by the compelling contents of online images, which represents one of the first attempts toward effective online mashup applications which connect a photo-sharing site with an advertising agency.
- 7
- Kernel-Based Linear Neighborhood Propagation for Semantic Video Annotation7
This approach combines the consistency assumption and the Local Linear Embedding (LLE) method in a nonlinear kernel-mapped space, which improves a recently proposed method Local Neighborhood Propagation (LNP) by tackling the limitation of its local linear assumption on the distribution of semantics.
- Tracking concept drifting with Gaussian mixture model7
The proposed GMM-based batch learning framework is effective and efficient for dealing with concept drifting in temporal sequences, such as video streams, over an extended period of time.
- Dual Dynamic Threshold Adjustment Strategy6
This work designs a static Asymmetric Sample Mining Strategy and its dynamic version, the Adaptive Tolerance ASMS, tailored for sample mining methods, and combines these two threshold adjustment algorithms to form the Dual Dynamic Threshold Adjustment Strategy (DDTAS).
- Panet: Perspective-Aware Network with Dynamic Receptive Fields and Self-Distilling Supervision for Crowd Counting6
This paper proposes a novel perspective-aware approach called PANet, which outperforms the state-of-the-art methods by a large margin and proposes the dynamic receptive fields (DRF) framework, which is able to adjust the receptive field by the dilated convolution parameters according to the input image.
- CoHOZ6
The ambition for solving the combination of these two problems to fulfill semantically recognizing the unknown classes detected in OSR by zero-shot prediction is taken and the Contrastive multimodal prompt tuning for Hierarchical Open-set Zero-shot recognition (CoHOZ) is proposed.
- Second-Order Camera-Aware Color Transformation for Cross-Domain Person Re-identification6
A Second-order Camera-aware Color Transformation (SCCT) that can operate on image level and align the second-order statistics of all the views of both source and target domain data with original ImageNet data statistics is proposed.
- Spatial likelihood voting with self-knowledge distillation for weakly supervised object detection6
In this framework, a spatial likelihood voting (SLV) module is introduced to converge region proposal localization without bounding box annotations and a self-knowledge distillation module is proposed to refine the feature representations of the given image.
- Challenges and Practices of Large Scale Visual Intelligence in the Real-World6
This talk will introduce current practices of the City Brain platform, as well as what can be done to achieve the goal and make it a reality, step by step, and detail how Luban was created and how it changed the world of 2D banner design by creating 50M banners a day.
- Exploiting On-Device Image Classification for Energy Efficiency in Ambient-Aware Systems6
This article uses a low-complexity feature-based image classifiers to filter out unnecessary frames from video and proposes a hierarchically pipelined hardware architecture for the image classifier, which can achieve minimum-energy operation at a frame rate of 12 fps, while consuming only 3 mJ of energy per frame.
- WonderWhat: real-time event determination from photos6
A system that provides users with information of the public events that they are attending by analyzing in real time their photos taken at the event, leveraging both spatio-temporal context and photo content is proposed.
- Community Discovery from Movie and Its Application to Poster Generation6
A new method to discover key roles and their relationship by treating a movie as a small community is proposed, and the automatic generation of video poster based on the discovered community is presented.
- Graph-cut based tag enrichment6
Experiments on Flickr dataset demonstrate the effectiveness of the proposedGraph cut based tag enrichment approach, which builds a graph for each image with its initial tags.
- Image Search Result Summarization with Informative Priors6
It is argued that the relevance and image quality are two important measures for a user friendly summarization since image search results are normally noisy with some low-quality images.
- Intention-based home video browsing6
The previous work is extended to build up a comprehensive scheme to mine the capture intention, and based on this scheme, a novel home video browsing system is proposed that offers users a more efficient way to search a given clip in a video than attention-based scheme.
- To mine capture intention of camcorder users6
This approach models the capture intention from a camcorder user's point of view, by investigating a set of effective intention oriented features, and concludes that the intention based approach is an effective complement to existing home video content analysis schemes.
- An Online Learning Framework for Sports Video View Classification6
An online learning framework for sports video structure analysis, using baseball as an example, in which only a very small number of pre-labeled training samples are required at initial stage, employs an optimal local positive model by sufficiently exploring the local statistic characteristics of the current under-test videos.
- 5
- Part-Aware Attention Network for Person Re-identification5
A novel Part-aware Attention Network (PAN) is proposed to connect part feature maps and middle-level features and achieves leading performance on Re-ID benchmarks Market1501, DukeMTMC, and CUHK03.
- Weakly Supervised Organ Localization with Attention Maps Regularized by Local Area Reconstruction5
The authors' weakly supervised organ localization network, namely OLNet, can generate high-resolution attention maps that preserve fine-detailed target anatomical structures and can provide promising localization results both in saliency map and semantic segmentation perspectives.
- 5
- 5
- 5
- 5
- Dynamic Video Collage5
This demo presents a video visualization technique named Dynamic Video Collage (DVC), which aims to mimic the storytelling of the video content by dynamically presenting a series of collages.
- 5
- 5
- 5
- 5
- 4
- Efficient Token Compression for Vision Transformer with Spatial Information Preserved4
This work proposes an efficient and hardware-compatible token compression method called Prune and Merge, which integrates token pruning and merging operations within transformer models to achieve layer-wise token compression.
- Hierarchical Graph Pattern Understanding for Zero-Shot VOS4
A new hierarchical graph neural network (GNN) architecture, dubbed hierarchical graph pattern understanding (HGPU), is proposed for zero-shot video object segmentation (ZS-VOS), inspired by the strong ability of GNNs in capturing structural relations.
- 4
- 4
- 4
- 4
- 4
- Mobile Cloud Visual Media Computing: From Interaction to Service4
This book explores the internet and mobile ecosystems which are powered by cloud computing, focusing on the intersection of mobile, cloud, visual and multimedia computing and is split into five clear parts.
- 4
- Towards next generation multimedia recommendation systems4
This course aims to provide a series of detailed review of state-of-the-art in multimedia recommendation, analyze key technical challenges in developing and evaluating next generation multimedia recommendation systems from different perspectives and give some predictions about the road lies ahead of us.
- A comprehensive representation scheme for video semantic ontology and its applications in semantic concept detection4
This paper presents a comprehensive representation scheme for video semantic ontology in which all the three components are well studied, and leverage LSCOM to construct the concept lexicon, describe concept property as the weights of different modalities which are obtained manually or by data-driven approach.
- ACM workshop on mobile cloud media computing4
The focus of this workshop is on exploring challenges and opportunities of intelligent multimedia technologies, applications and systems on mobile devices, especially when a media cloud computing platform can be appropriately leveraged.
- 4
- Color-Structured Image Search4
A novel interactive image ranking technique, color-structured image search, by exploiting color spatial relation information, is proposed and demonstrated in image research results refinement and a convenient interactive interface is presented to allow users to specify target color structure flexibly.
- Concept representation based video indexing4
This poster introduces a novel concept-based video indexing approach that is developed based on a rich set of base concepts, of which the models are available, and concludes that the sparse method will lead to much better performance.
- Introduction to computer vision and image understanding the special issue on video analysis4
This special issue of the Computer Vision and Image Understanding offers a venue to present innovative approaches in video analysis, which have been changing the authors' everyday life dramatically over the last few years, and fills the need of a comprehensive overview of the new approaches.
- Optimizing Training Set Construction for Video Semantic Classification4
It is argued that a training set which includes most temporal and spatial distribution information of the whole data will achieve a good performance even if the size of training set is limited, and four metrics for constructing/selecting an optimal training set are proposed, including salience, temporal disperseiveness, spatial dispersiveness, and diversity.
- 4
- 4
- Natural video browsing4
In this demonstration, a novel system, Video Booklet, which enables nature personal video browsing and searching, and builds a seamless bridge between digital videos and analog albums.
- 4
- DREAM: Decoupled Discriminative Learning with Bigraph-aware Alignment for Semi-supervised 2D-3D Cross-modal Retrieval3
The core of the DREAM is to decouple the label prediction and reliability measurement processes to reduce overconfident samples in discriminative learning and build a bigraph graph that connects cross-modal similar examples and then conduct learning to cluster with most edges kept for alignment.
- 3
- 3
- 3
- Searching for images by video3
A framework in which the problem of reliability of QBE-based image retrieval is explicitly addressed and the rich information contained therein can be exploited to discover the proper query representation and to improve the relevance of the retrieved results is proposed.
- Clip-based hierarchical representation for near-duplicate video detection3
This paper proposes to use multi-probe locality sensitive hashing (MPLSH) to index the video clips for fast similarity search and high recall, and applies a similarity voting based on video clip signatures to refine the search process.
- 3
- Interactive multimedia computing3
This special issue is introducing novel research work on interactive multimedia computing, including game-based image annotation, interactive TV, interactive cartoon synthesis, and so on, and discusses the security and privacy issues in the context of interactive TV.
- 3
- 3
- 3
- An Efficient Automatic Video Shot Size Annotation Scheme3
An efficient learning scheme for automatic annotation of video shot size by using a more general low- and mid- level feature set based on two complementary feature sets is presented.
- 3
- 3
- Online Play Segmentation for Broadcasted American Football TV Programs3
An online adaptive field-color model is employed to deal with feature variations which may caused by court configurations and time-varying lighting conditions and a temporal pattern based model is constructed to capture intrinsic features of football plays, which considerably increased the performance of the proposed scheme.
- 2
- 2
- Deep Graph Clustering with Disentangled Representation Learning2
A novel deep graph clustering framework named DisenCluster is proposed, which learns disentangled representations to simultaneously consider node separation results from diverse perspectives and is formulated as a continuous relaxation of the high-order graph cut problem and optimize the objective to obtain effective cluster assignments.
- 2
- Towards Graph Prompt Learning: A Survey and Beyond2
Through this extensive review, this survey provides a foundational understanding of graph prompt learning, aiming to impact not only the graph mining community but also the broader Artificial General Intelligence (AGI) community.
- 2
- 2
- 2
- Optimizing traffic efficiency via a reinforcement learning approach based on time allocation2
A novel method is proposed that can naturally provide a countdown function by adopting a new action design that achieves control in the manner of time allocation, and the model figures out the duration of each phase at the beginning of every signal cycle, which is more practical for real-world traffic applications.
- Proposal-Level Unsupervised Domain Adaptation for Open World Unbiased Detector2
This work aims to build an unbiased foreground predictor by re-formulating the task under Unsupervised Domain Adaptation, where the current biased predictor helps form the domains: the seen object locations and confident background locations as the source domain, and the rest ambiguous ones as the target domain.
- Dynamic supervisor for cross-dataset object detection2
It is pointed out that it is not enough to generate high-quality annotations using a single model, which only looks once for annotations, and a dynamic supervisor framework that updates the annotations multiple times through multiple-updated submodels trained using hard and soft labels is proposed.
- Spatiotemporal Multi-Task Network for Human Activity Understanding2
A spatiotemporal, multi-task, 3D deep convolutional neural network to detect actions in untrimmed videos and introduces a novel video representation, interlaced images, as an additional network input stream to better utilize the rich motion information in videos.
- Multifold Concept Relationships Metrics2
This paper first defines five specific relationships between concepts and leverage them to annotate the images collected from commercial search engine and extract some conceptual features in textual and visual domain to train the concept model.
- 2
- StoryImaging2
The StoryImaging system is developed to illustrate a textual story with both images harvested from the Web and synthesized speech to improve information comprehension and provide engaging and explorative experiences for users by closely bridging textual and visual modalities.
- Visual Content Identification and Search2
This special issue contains several research articles covering a diverse range of topics in visual content identification and search, concerning image and video content, by visual similarities.
- Internet multimedia advertising2
This tutorial aims at bringing together recent insights from the research on multimedia advertising that addresses the theoretical fundamentals, solution concepts, and the issues related to the development of modern multimedia advertising schemes.
- Learning to Rank with Graph Consistency2
A Content-Aware Ranking model based on “learning to rank” framework, in which textual and visual information are simultaneously leveraged in the ranking learning process is proposed, which significantly outperforms the state-of-the-art ranking and reranking methods.
- Video Content Structure2
The present invention provides an antistatic polyester film having an antISTatic layer formed of an anti-freeze agent composed mainly of a polymer having a recurring unit of a structure expressed by the following formula.
- NLVS2
This video demonstration presents a new system for video summarization, called ``Near-Lossless Video Summarization" (NLVS) to tackle with the challenges of storage and indexing brought by the current boom of videos for existing online video services.
- Gamesense2
A novel game-like advertising system called GameSense, which is driven by the compelling contents of online images, is presented, able to provide viewers rich experience and thus promote the embedded ads to provide more effective advertising.
- 2
- 2
- Dual Data-centric Separation with Circular Mixup for Noise-resistant Time Series Learning1
This paper proposes a novel data-centric approach named Dual Data-centric Separation with Circular Mixup (DREAM) to explore and exploit the noisy data from dual data-centric views for reduced overfitting of noise-resistant time series learning.
- 1
- 1
- DisCo: Diffusion-guided Unbiased Discriminative Learning for Unsupervised Graph Domain Adaptation1
The core of the DisCo is to leverage both feature disentanglement and cross-domain diffusion signals to remove the potential biases for target graphs, and first utilize adversarial feature disentanglement to extract causal features that are orthogonal to domain biases.
- Aligning LLM Uncertainty with Human Disagreement in Subjectivity Analysis1
Experiments on three subjectivity analysis tasks show that DPUA preserves task performance while better aligning model uncertainty with human disagreement, mitigating overconfidence on boundary samples, and improving out-of-distribution generalization.
- 1
- 1
- SPOT: Spectral Optimal Transport for Graph Domain Generalization1
This work formulate the structural augmentation as an optimal transport problem to retain low-frequency key knowledge and solve the problem using Sinkhorn-Knopp algorithm, and incorporates an adaptive perturbation strategy to deep features, where the direction of the additive noise is determined by the homophily degrees to maintain semantic properties.
- 1
- 1
- 1
- 1
- Delving into Details: Synopsis-to-Detail Networks for Video Recognition1
The proposed synopsis-to-detail networks enable us to train models directly from scratch in an end-to-end manner and to investigate various architectures for synopsis/detail recognition.
- 1
- Ranking Model Adaptation for1
A regularization based algorithm called ranking adaptation SVM (RA-SVM), through which an existing ranking model is adapted to a new domain, so that the amount of labeled data and the training cost is reduced while the performance is still guaranteed.
- Query-driven iterated neighborhood graph search for scalable visual indexing1
This paper investigates a simple but very effective approach, neighborhood graph (NG) search, which conducts the local search by expanding neighborhoods with a best-first manner, and proposes a query-driven iterated neighborhood graph search approach to improve the performance.
- Special Issue on Subspace and Manifold Learning for Image and Video Indexing and Search1
The four papers in this special issue focus on the novel design and methodology of subspace and manifold learning for image and video indexing and search.
- Typicality ranking: beyond accuracy for video semantic annotation1
Three pre-processing approaches are proposed to relax the labels of the training data to real-valued typicality scores, and a novel criterion, Average Typicality Precision (ATP), is proposed to replace the frequently used one, Average Precision (AP), for evaluating the performance of video typicality ranking algorithms.
- Elements of Visual Concept Analysis1
In order to facilitate the following discussion on concept distance measurements, the generative models, such as bag-of-words model, 2D hidden markov model, visual language model are mainly emphasized.
- TapTell1
This demonstration presents a mobile-based visual recognition and recommendation application on Windows Phone 7 called TapTell, which discovers and understands users' visual intents via a circle based natural user interaction called "O" gestures and recommends relevant entities utilizing contextual information.
- 1
- Web-scale image search by color sketch1
This demonstration presents an image search system, which enables consumers to find images on the requirement of how the colors are spatially distributed, and is the first one to scale up to Web-scale images.
- Semi-automatic photo clustering with distance metric learning1
This paper proposes a semi-automatic photo clustering scheme that attempts to improve clustering performance with users' interactions, and it is better than pure manual adjustment approach by exploring distance metric learning.
- ImageSense1
This demonstration presents an innovative contextual advertising platform for online image service, called ImageSense, which aims to embed more contextually relevant ads at less intrusive positions within each suitable image.
- 1
- 1
- 1
- –
- –
- –
- –
- PROBE: VLM-Guided Discrete Structural Reconfiguration for Customized and Efficient Video Retrieval–
An efficient VLM-guided video retrieval paradigm PROBE is proposed that leverages a frozen vision–language model (VLM) to inject rich semantic structure into a reusable hash index and introduces criterion-induced subspace regularization that transfers semantic geometry into criterion-conditioned hash subspaces while preserving local relational consistency.
- –
- –
- –
- –
- Anti-Collapse Loss for Deep Metric Learning Based on Coding Rate Metric–
This work proposes a novel loss function called Anti-Collapse Loss that promotes the sparseness of feature clusters in the embedding space to prevent collapse by maximizing the average coding rate of sample features or class proxies and integrates with pair-based and proxy-based methods, resulting in notable performance improvement.
- –
- –
- –
- –
- –
- –
- –
- Stable Learning via Causality-based Feature Rectification.–
A novel Causality-based Feature Rectification (CFR) method to address the model misspecification problem under agnostic distribution shift by using a weight matrix to rectify features and can not only improve the stability of linear models, but also deep-learning based models.
- –
- Partial Feature Decorrelation for Non-I.I.D Image classification–
A novel Partial Feature Decorrelation Learning (PFDL) Algorithm is presented, which jointly optimizes a feature decomposition network and the target image classification model, and can improve the backbone model's accuracy in non-i.i.d image classification datasets.
- –
- Stable Learning via Self-supervised Invariant Risk Minimization–
This work proposes a novel Self-supervised Invariant Risk Minimization method based on the fact that the real causality connections between features are consistent no matter how the environment changes, which outperforms previous state-of-the-art stable learning methods.
- –
- Cloud-Based Mobile Experience Sharing Through Automatic Multimedia Blogging–
This paper will propose a mobile + cloud system enabling rapid and near-realtime experience sharing through automatic blogging and microblogging, which are based on multimodal media content analyses and syntheses.
- –
- –
- Special Section on Learning from Multiple Evidences for Large Scale Multimedia Analysis: editorial–
This special issue is presenting the latest research on combing multiple evidences for multimedia analysis, and proposes a novel method to automatically retrieve the video shots of that action from Internet by jointly exploiting the metadata and visual features of web videos.
- –
- Special section on learning from multiple evidences for large scale multimedia analysis–
This special issue is presenting the latest research on combing multiple evidences for multimedia analysis, and proposes a novel method to automatically retrieve the video shots of that action from Internet by jointly exploiting the metadata and visual features of web videos.
- Introduction to the Special Section on Intelligent Multimedia Systems and Technology Part II–
This section seeks effective machine-learning and data-mining algorithms, frameworks, systems, and implementations that particularly work on multimedia data (including image, video, and audio, which may also be associated with textual information).
- –
- Message from ICME'12 Technical Program Chairs–
The ICME 2012 daily technical program has been designed in such a way that ICME audience will not miss the keynotes, overview talks and the plenary session, and can always find a high quality talk of interest at the parallel oral sessions.
- In-Image Advertising–
Two implications of the theory are that image advertising campaigns are necessarily massive, and that the targeted audience for the advertising campaign is necessarily larger than the targeted group of buyers of the advertised product.
- Introduction to the special issue on intelligent multimedia systems and technology–
This special issue seeks effective machine-learning and data-mining algorithms, frameworks, systems, and implementations that particularly work on multimedia data (including image, video, and audio, which may also be associated with textual information).
- Melog–
Melog, a "mobile + cloud" multimedia system enabling efficient and near-realtime experience sharing through automatic blogging and micro-blogging, which are based on multi-modal media content analyses and syntheses is demonstrated.
- –
- Pairwise Constrained Semi-supervised Maximum Margin Clustering–
A semi-supervised maximum margin clustering algorithm that incorporates additional pairwise constraints and a pairwise loss function is introduced into the clustering objective function which effectively penalizes the violation of the given constraints.
- –
- A Unifying Multi-Label Temporal Kernel Machine with Its Application to Video Annotation–
This paper proposes a new paradigm of the video annotation method that simultaneously annotates the concepts as well as model correlations between them in one step by the proposed Correlative Multi-Label (CML) method and extends the proposed method to exploit the rich temporal information in the videos.
- Content‐Based Multimedia Retrieval–
Both classic CBMR and the current mainstream are introduced, where QBE is the most typical scenario for multimedia retrieval before the year of 2000, whereas query by text and by combination of text and examples have become two new mainstream scenarios thereafter.
- LazyCap - Template-Based Media Capturing–
This paper proposes a novel scheme, Template-Based Media Capturing, which provides a uniform scheme to collect extra information during the process of media capturing in an easy and convenient manner, and remarkably facilitates further personal media applications such as management, authoring and sharing.
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-14. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.