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
- Xiuqing WuSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works588 from public data
- NUS-WIDE3,409
The benchmark results indicate that it is possible to learn effective models from sufficiently large image dataset to facilitate general image retrieval and four research issues on web image annotation and retrieval are identified.
- Richer Convolutional Features for Edge Detection1,240
RCF encapsulates all convolutional features into more discriminative representation, which makes good usage of rich feature hierarchies, and is amenable to training via backpropagation, and achieves state-of-the-art performance on several available datasets.
- 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.
- 458
- Generalized Focal Loss V2: Learning Reliable Localization Quality Estimation for Dense Object Detection450
This paper explores a completely novel and different perspective to perform LQE – based on the learned distributions of the four parameters of the bounding box – and develops a considerably lightweight Distribution-Guided Quality Predictor (DGQP) for reliable LqE based on GFLV1, thus producing GFLv2.
- 427
- 404
- Outer Product-based Neural Collaborative Filtering363
Extensive experiments demonstrate the effectiveness of the proposed ONCF framework, in particular, the positive effect of using outer product to model the correlations between embedding dimensions in the low level of multi-layer neural recommender model.
- 360
- 360
- Centralized Feature Pyramid for Object Detection342
A Centralized Feature Pyramid (CFP) network for object detection, which is based on a globally explicit centralized feature regulation, which not only has the ability to capture the global long-range dependencies but also efficiently obtain an all-round yet discriminative feature representation.
- CLIP-Driven Fine-Grained Text-Image Person Re-Identification329
A CLIP-driven Fine-grained information excavation framework (CFine) to fully utilize the powerful knowledge of CLIP for TIReID and proposes a multi-level global feature learning (MGF) module that fully mines the discriminative local information within each modality, thereby emphasizing identity-related discrim inative clues through enhanced interaction between global image (text) and informative local patches (words).
- 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.
- Generalized Nonconvex Nonsmooth Low-Rank Minimization293
In theory, it is proved that IRNN decreases the objective function value monotonically, and any limit point is a stationary point, which enhances the low-rank matrix recovery compared with state-of-the-art convex algorithms.
- Spatially-Adaptive Feature Modulation for Efficient Image Super-Resolution289
This work proposes a spatially-adaptive feature modulation (SAFM) mechanism for efficient SR design that is 3× smaller than state-of-the-art efficient SR methods, e.g., IMDN, and yields comparable performance with much less memory usage.
- 289
- Generalized Focal Loss: Learning Qualified and Distributed Bounding Boxes for Dense Object Detection271
- Image annotation by k NN-sparse graph-based label propagation over noisily tagged web images269
This article exploits the problem of annotating a large-scale image corpus by label propagation over noisily tagged web images by proposing a novel kNN-sparse graph-based semi-supervised learning approach for harnessing the labeled and unlabeled data simultaneously.
- 258
- 257
- 250
- 243
- 238
- 238
- 235
- Online topic-aware influence maximization234
This work proposes a faster topic-sample-based algorithm with e · (1 − 1/e) approximation ratio for any e ∈ (0, 1], which materializes the influence spread of some topic-distribution samples and utilizes the materialized information to avoid computing the actual influence of users with small influences.
- 231
- 227
- Learning Dual Convolutional Neural Networks for Low-Level Vision221
The proposed DualCNN consists of two parallel branches, which respectively recovers the structures and details in an end-to-end manner, and can generate the target signals according to the formation model for each particular application.
- Weakly-Shared Deep Transfer Networks for Heterogeneous-Domain Knowledge Propagation214
This paper develops a novel deep network structure, capable of transferring labeling information across heterogeneous domains, especially from text domain to image domain, and presents a novel architecture of DTNs to translate cross-domain information from text to image.
- Hierarchical Long Short-Term Concurrent Memory for Human Interaction Recognition213
A novel Hierarchical Long Short-Term Concurrent Memory (H-LSTCM) is proposed to model the long-term inter-related dynamics among a group of persons for recognizing human interactions by comparing against baseline and state-of-the-art methods.
- Inferring semantic concepts from community-contributed images and noisy tags212
This paper proposes a novel sparse graph-based semi-supervised learning approach for harnessing the labeled and unlabeled data simultaneously, and constructs an informative compact concept space with small semantic gap to infer the semantic concepts in this space to bridge the semantic gap.
- 211
- Adversarial Training Towards Robust Multimedia Recommender System208
This paper proposes a novel solution named Adversarial Multimedia Recommendation (AMR), which can lead to a more robust multimedia recommender model by using adversarial learning, to train the model to defend an adversary, which adds perturbations to the target image with the purpose of decreasing the model's accuracy.
- VALOR: Vision-Audio-Language Omni-Perception Pretraining Model and Dataset207
Extensive experiments show that VALOR can learn strong multimodal correlations and generalize to various downstream tasks (e.g., retrieval, captioning, and question answering) with different input modalities (e.g., vision-language, audio-language, and audiovisual-language).
- 202
- 197
- 194
- Weakly-supervised Semantic Guided Hashing for Social Image Retrieval192
This work proposes a novel Semantic Guided Hashing method coupled with binary matrix factorization to perform more effective nearest neighbor image search by simultaneously exploring the weakly-supervised rich community-contributed information and the underlying data structures.
- Physics-Based Generative Adversarial Models for Image Restoration and Beyond191
An algorithm to directly solve numerous image restoration problems (e.g., image deblurring, image dehazing, and image deraining) by generative models with adversarial learning within the GAN framework.
- 189
- 183
- Cascaded Deep Video Deblurring Using Temporal Sharpness Prior182
A temporal sharpness prior to constrain the deep CNN model to help the latent frame restoration and it is shown that exploring the domain knowledge of video deblurring is able to make the deepCNN model more compact and efficient.
- IMAGPose: A Unified Conditional Framework for Pose-Guided Person Generation173
The proposed IMAGPose is a unified conditional framework for pose-guided image generation, which incorporates three pivotal modules: a feature-level conditioning (FLC) module, an image-level conditioning (ILC) module, and a cross-view attention (CVA) module that work together to unify the task of person image generation under various user scenarios.
- BlockMix: Meta Regularization and Self-Calibrated Inference for Metric-Based Meta-Learning166
This work proposes new mechanisms to learn generalized and discriminative feature embeddings as well as improve the robustness of classifiers against prediction corruptions for meta-learning by integrating interpolation on the images and labels within metric learning.
- 163
- 160
- 159
- 159
- Image-Specific Information Suppression and Implicit Local Alignment for Text-Based Person Search158
An efficient joint multilevel alignment network (MANet) for TBPS is proposed, which can learn aligned image/text feature representations between modalities at multiple levels, and realize fast and effective person search.
- Boosting Few-Shot Fine-Grained Recognition With Background Suppression and Foreground Alignment155
A two-stage background suppression and foreground alignment framework is introduced to generate a foreground mask for localization to weaken background disturbance and enhance dominative foreground objects and a novel L2L similarity metric is presented to further measure the local similarity between a pair of aligned spatial features in the embedding space.
- Personalized Age Progression with Aging Dictionary151
The advantages of the proposed solution over other state-of-the-arts in term of personalized aging progression, as well as the performance gain for cross-age face verification by synthesizing aging faces are demonstrated.
- Adaptive Context Network for Scene Parsing149
This paper proposes an Adaptive Context Network (ACNet) to capture the pixel-aware contexts by a competitive fusion of global context and local context according to different per-pixel demands.
- Generalized Deep Transfer Networks for Knowledge Propagation in Heterogeneous Domains144
A novel generalized deep transfer networks (DTNs) capable of transferring label information across heterogeneous domains, textual domain to visual domain, and to share the labels between two domains are proposed, able to generate domain-specific and shared interdomain features.
- 144
- Accurate and Efficient Stereo Matching via Attention Concatenation Volume143
A novel cost volume construction method, named attention concatenation volume (ACV), which generates attention weights from correlation clues to suppress redundant information and enhance matching-related information in the concatenation volume.
- Participation-Contributed Temporal Dynamic Model for Group Activity Recognition140
A novel Participation-Contributed Temporal Dynamic Model (PC-TDM) to recognize group activity, which mainly consists of a "One" network and a " one to Key" network, which aims at modeling the individual dynamic of each person.
- 138
- Ecological dynamics of the gut microbiome in response to dietary fiber136
It was found that inulin produced a generally rapid response followed by gradual stabilization to new equilibria, and those dynamics were baseline-dependent, which emphasizes the importance of ecological modeling to understand microbiome responses to dietary changes and the need for personalized interventions.
- M3Net: Multi-view Encoding, Matching, and Fusion for Few-shot Fine-grained Action Recognition134
M3Net is proposed, a matching-based framework for FS-FG action recognition, which incorporates multi-view encoding, multi- view matching, and multi-View fusion to facilitate embedding encoding, similarity matching,and decision making across multiple viewpoints.
- 130
- 128
- 126
- 125
- Modeling Multimodal Clues in a Hybrid Deep Learning Framework for Video Classification123
A novel hybrid deep learning framework is introduced that integrates useful clues from multiple modalities, including static spatial appearance information, motion patterns within a short time window, audio information, as well as long-range temporal dynamics.
- Social Adaptive Module for Weakly-Supervised Group Activity Recognition121
This paper presents a new task named weakly-supervised group activity recognition (GAR) which differs from conventional GAR tasks in that only video-level labels are available, yet the important persons within each frame are not provided even in the training data.
- W2Go120
A novel Automatic Landmark Ranking (ALR) method is proposed by utilizing the tag and geo-tag information of photos in Flickr and user knowledge from Yahoo Travel Guide, which can automatically recognize and rank the landmarks for travellers.
- An efficient sparse metric learning in high-dimensional space via l 1 -penalized log-determinant regularization119
An efficient sparse metric learning algorithm in high dimensional space via an l1-penalized log-determinant regularization that can be efficiently optimized in a block coordinate descent fashion which is much faster than the standard semi-definite programming which has been widely adopted in many other advanced distance learning algorithms.
- Deep Semantic Multimodal Hashing Network for Scalable Image-Text and Video-Text Retrievals118
The proposed DSMHN method is a generic and scalable deep hashing framework for both image-text and video-text retrievals, which can be flexibly integrated with different types of loss functions simultaneously.
- 117
- Reactive force-field molecular dynamics study on graphene oxide reinforced cement composite: functional group de-protonation, interfacial bonding and strengthening mechanism116
Reactive force-field molecular dynamics was utilized to investigate the structure, reactivity and interfacial bonding of calcium silicate hydrate/GO nanocomposite functionalized by hydroxyl, epoxy, carboxyl, and sulfonic groups and revealed that C-S-H reinforced with GO-COOH and GO-OH had better interfacial cohesive strength and ductility than that observed under tensile loading.
- 115
- 113
- Personalized Age Progression with Bi-Level Aging Dictionary Learning111
This work aims to automatically render aging faces in a personalized way by proposing a novel Bi-level Dictionary Learning based Personalized Age Progression (BDL-PAP) method, formulated to learn the aging dictionaries based on face pairs from neighboring age groups.
- Constructing a Nonnegative Low-Rank and Sparse Graph With Data-Adaptive Features110
Extensive NNLRS experiments demonstrate that the proposed method outperforms the state-of-the-art graph construction method by a large margin for both semisupervised classification and discriminative analysis, which verifies the effectiveness of this proposed method.
- 109
- 106
- 106
- Beyond search106
A novel scheme that is able to summarize the content of video search results by mining and threading “key” shots, such that users can get an overview of main content of these videos at a glance is presented.
- Pedestrian-specific Bipartite-aware Similarity Learning for Text-based Person Retrieval105
A novel Pedestrian-specific Bipartite-aware Similarity Learning (PBSL) framework is proposed that efficiently reveals the plausible and credible levels of contribution of pedestrian-specific mismatched and matched region-word pairs towards overall similarity.
- 98
- Deep Blind Video Super-resolution98
An effective blind video SR algorithm based on deep convolutional neural networks (CNNs) and an effective sharp feature exploration method which first extracts sharp features from restored intermediate latent frames and then uses a transformation operation based on the extracted sharp features and warped features from LR inputs to generate better features for HR video restoration.
- 96
- 95
- Deep Ordinal Hashing With Spatial Attention95
A novel deep ordinal hashing (DOH) method is proposed, which learns ordinal representations to generate ranking-based hash codes by leveraging the ranking structure of feature space from both local and global views.
- 95
- Correntropy Induced L2 Graph for Robust Subspace Clustering95
This paper proposes a new robust clustering method by using the correntropy induced metric, which is robust for handling the non-Gaussian and impulsive noises and extends the method for dealing with the data with outlier rows/features.
- Efficient Visual State Space Model for Image Deblurring92
This paper proposes a simple yet effective visual state space model (EVSSM) for image deblurring, leveraging the benefits of state space models (SSMs) for visual data, and proposes an efficient discriminative frequency domain-based feedforward network (EDFFN), which can effectively estimate useful frequency information for latent clear image restoration.
- 92
- The Ninth NTIRE 2024 Efficient Super-Resolution Challenge Report91
This paper provides a comprehensive review of the NTIRE 2024 challenge, focusing on efficient single-image super-resolution (ESR) solutions and their outcomes, and gauge the state-of-the-art in efficient single-image super-resolution.
- Erasing, Transforming, and Noising Defense Network for Occluded Person Re-Identification91
A simple yet effective framework which treats occlusion as a noise disturbance and solves occluded person re- ID from the perspective of adversarial defense, and designs the first GAN-based adversarial defense paradigm for occluded person re- ID.
- Supervised deep hashing for scalable face image retrieval90
This work proposes a novel supervised hashing method for scalable face image retrieval, i.e., Deep Hashing based on Classification and Quantization errors (DHCQ), by simultaneously learning feature representations of images, hash codes and classifiers.
- 89
- Label to region by bi-layer sparsity priors89
This work proposes a bi-layer sparse coding formulation for uncovering how an image or semantic region can be robustly reconstructed from the over-segmented image patches of an image set and investigates how to automatically reassign the manually annotated labels at the image-level to those contextually derived semantic regions.
- How to Learn Item Representation for Cold-Start Multimedia Recommendation?88
A Multi-Task Pairwise Ranking (MTPR) framework for model training is proposed, which enforces the observed interactions ranking higher than the unobserved ones even if the item is assumed to be cold.
- Beyond Object Proposals: Random Crop Pooling for Multi-Label Image Recognition84
This paper proposes an object-proposal-free framework for multi-label image recognition: random crop pooling (RCP), which performs stochastic scaling and cropping over images before feeding them to a standard convolutional neural network, which works quite well with a max-pooling operation for recognizing the complex contents of multi- label images.
- DLGSANet: Lightweight Dynamic Local and Global Self-Attention Network for Image Super-Resolution83
A sparse global self-attention (SparseGSA) module is developed to select the most useful similarity values so that the most useful global features can be better utilized for image reconstruction.
- Joint Label Prediction Based Semi-Supervised Adaptive Concept Factorization for Robust Data Representation82
A joint label prediction based Robust Semi-Supervised Adaptive Concept Factorization (RS2ACF) framework that incorporates the joint neighborhood reconstruction error over the new representations and predicted labels of both labeled and unlabeled data, so the manifold structures can be preserved explicitly and adaptively in the representation space and label space at the same time.
- Different Users, Different Opinions82
Inspired by recent studies in predicting result relevance based on mouse movement patterns (namely motifs), this work proposes to estimate the utilities of search results and the efforts in search sessions with motifs extracted from mouse movement data on search result pages (SERPs).
- 81
- 81
- Social Anchor-Unit Graph Regularized Tensor Completion for Large-Scale Image Retagging81
This work proposes a novel Social anchor-Unit GrAph Regularized Tensor Completion (SUGAR-TC) method to efficiently refine the tags of social images, which is insensitive to the scale of data.
- 81
- 80
- 79
- 79
- 78
- 77
- Fast Matrix Factorization With Nonuniform Weights on Missing Data77
A fast learning method, for which the time complexity is determined by the number of observed entries in the data matrix rather than the matrix size, and the key idea is truncated singular value decomposition on the weight matrix to get a more compact representation of the weights.
- WLBP: Weber local binary pattern for local image description77
A local descriptor, called Weber Local Binary Pattern (WLBP 1), which effectively combines the advantages of WLD and LBP, which is evaluated on the face recognition problem under different challenges and demonstrates that WLBP is a powerful texture descriptor.
- 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.
- Long-Term TalkingFace Generation via Motion-Prior Conditional Diffusion Model75
The MCDM model is introduced, which utilizes both archived and current clip motion priors to enhance motion prediction and ensure temporal consistency in TalkingFace generation.
- Supervised Quantization for Similarity Search74
This paper presents a compact coding approach, supervised quantization, that simultaneously learns feature selection that linearly transforms the database points into a low-dimensional discriminative subspace, and quantizes the data points in the transformed space.
- 73
- 73
- Towards Unified Deep Image Deraining: A Survey and a New Benchmark72
An online platform is built to provide the off-the-shelf toolkit, involving the large-scale performance evaluation of existing image deraining methods, and a new high-quality benchmark named HQ-RAIN is constructed to conduct extensive evaluations.
- 72
- 72
- 70
- 70
- 69
- 69
- Merging Context Clustering With Visual State Space Models for Medical Image Segmentation68
This work introduces a simple yet effective method named context clustering ViM (CCViM), which incorporates a context clustering module within the existing ViM models to segment image tokens into distinct windows for adaptable local clustering.
- 68
- 68
- 67
- 65
- Deep Learning Driven Visual Path Prediction From a Single Image64
A deep learning framework is proposed, which simultaneously performs deep feature learning for visual representation in conjunction with spatiotemporal context modeling and a unified path-planning scheme is proposed to make accurate path prediction based on the analytic results returned by the deep context models.
- 63
- Learning Comprehensive Representations with Richer Self for Text-to-Image Person Re-Identification62
This work proposes a simple yet effective framework, called LCR2S, for modeling many-to-many correspondences of the same identity by learning comprehensive representations for both modalities from a novel perspective and distills the knowledge learned by the "richer" model into a lightweight model for inference with a single image/text as input.
- 62
- Invariant Representation Learning for Multimedia Recommendation62
An Invariant Representation Learning Framework (InvRL) is proposed to alleviate the impact of the spurious correlations in multimedia recommendation, and two iteratively executed modules to cluster user-item interactions and learn invariant representations are proposed.
- 62
- 62
- Living fabrication of functional semi-interpenetrating polymeric materials61
The engineering of living materials consisting of semi-interpenetrating polymer networks (sIPN), which serves the dual purposes of enhancing the mechanical property of the living materials and anchoring effector proteins for diverse applications, is demonstrated.
- 60
- 60
- 59
- 59
- 59
- Dual Convolutional Neural Networks for Low-Level Vision59
Experimental results show that the DualCNN can be effectively applied to numerous low-level vision tasks with favorable performance against the state-of-the-art methods that have been specially designed for each individual task.
- Multimodal Attention with Image Text Spatial Relationship for OCR-Based Image Captioning58
A novel design is presented - Multimodal Attention Captioner with OCR Spatial Relationship (dubbed as MMA-SR) architecture, which manages information from different modalities with a multimodal attention network and explores spatial relationships between text tokens for OCR-based image captioning.
- Show, Reward and Tell: Automatic Generation of Narrative Paragraph From Photo Stream by Adversarial Training58
This paper proposes a sequence-to-sequence modeling approach with reinforcement learning and adversarial training for the creation of narrative paragraph for a photo stream, and proposes a hierarchical recurrent neural network as story generator, optimized by reinforcement learning with rewards.
- Multimedia News Summarization in Search58
A novel approach of multimedia news summarization for searching results on the Internet is developed, which uncovers the underlying topics among query-related news information and threads the news events within each topic to generate a query- related brief overview.
- 58
- 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.
- 57
- Discriminative Deep Hashing for Scalable Face Image Retrieval57
A new Discriminative Deep Hashing network to learn discriminative and compact hash codes for large-scale face image retrieval and achieves superior performance compared with some state-of-the-art hashing methods.
- Topology preserving hashing for similarity search57
Topology Preserving Hashing is proposed, a novel hashing method that is distinct from prior works by preserving the neighborhood rankings of data points in Hamming space and is formulated as a generalized eigendecomposition problem with closed form solutions.
- 56
- FoundIR: Unleashing Million-Scale Training Data to Advance Foundation Models for Image Restoration54
A diffusion-based generalist model is utilized to remove degradations by learning the degradation-agnostic common representations from diverse inputs, where an incremental learning strategy is adopted to better guide model training.
- Context Disentangling and Prototype Inheriting for Robust Visual Grounding54
A novel framework with context disentangling and prototype inheriting for robust visual grounding to handle both standard and open-vocabulary scenes is proposed and performance comparisons indicate that the method outperforms the state-of-the-art methods in both scenarios.
- 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.
- 53
- Semi-supervised local feature selection for data classification53
This paper proposes a new semi-supervised local feature selection method (S2LFS) allowing to select different feature subsets for different classes by learning the importance of features considering each class separately.
- 52
- 52
- Spatial Structure Constraints for Weakly Supervised Semantic Segmentation51
A CAM-driven reconstruction module to directly reconstruct the input image from deep CAM features, which constrains the diffusion of last-layer object attention by preserving the coarse spatial structure of the image content, and an activation self-modulation module to refine CAMs with finer spatial structure details by enhancing regional consistency are proposed.
- 51
- Event driven summarization for web videos51
This paper presents a novel solution by mining and threading "key" shots, which can provide an overview of main contents of videos at a glance, by summarizing a large set of diverse videos by utilizing the way of dynamic video skimming.
- 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.
- Multi-Granularity Denoising and Bidirectional Alignment for Weakly Supervised Semantic Segmentation50
This work proposes an end-to-end multi-granularity denoising and bidirectional alignment model, to alleviate the noisy label and multi-class generalization issues, and proposes the online noise filtering and progressive noise detection modules to tackle image-level and pixel-level noise.
- 49
- 48
- 48
- Triplet Contrastive Representation Learning for Unsupervised Vehicle Re-Identification47
A triplet contrastive representation learning (TCRL) framework, which leverages cluster features to bridge the part features and global features for unsupervised vehicle re-identification and outperforms many state-of-the-art unsupervised vehicle re-identification approaches.
- 47
- 47
- 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.
- Convolutional‐capsule network for gastrointestinal endoscopy image classification45
The core of the method is a lesion‐aware CNN feature extraction module which can encode sufficiently detailed information of lesions in midlevel CNN features and in turn enable the subsequent capsule classification network to effectively learn deformation‐invariant relationships between image entities.
- Facial Age Synthesis With Label Distribution-Guided Generative Adversarial Network45
This paper proposes label distribution-guided generative adversarial network (ldGAN), where each sample is associated with an age label distribution rather than a single age group, and demonstrates the effectiveness of ldGAN in both learning short-term aging patterns and coping with the lack of training data.
- 44
- 44
- 42
- 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.
- 41
- 41
- 41
- 41
- 41
- 41
- 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.
- View-based 3D model retrieval with probabilistic graph model41
This work proposes a view-based 3D model retrieval algorithm using probabilistic graph model, and a solution to estimate the conditional probabilities of the models in the database given the query model.
- BiSTNet: Semantic Image Prior Guided Bidirectional Temporal Feature Fusion for Deep Exemplar-Based Video Colorization40
A BiSTNet to explore colors of exemplars and utilize them to help video colorization by a bidirectional temporal feature fusion with the guidance of semantic image prior is presented.
- 40
- 40
- 39
- The Tenth NTIRE 2025 Efficient Super-Resolution Challenge Report39
This paper presents a comprehensive review of the NTIRE 2025 Challenge on Single-Image Efficient Super-Resolution, emphasizing groundbreaking advancements in state-of-the-art single-image ESR techniques.
- 39
- 39
- Causal Inference with Knowledge Distilling and Curriculum Learning for Unbiased VQA39
This work proposes a two-stage unbiased VQA approach that addresses the unbiased issue from a causal perspective and demonstrates the superior performance of the proposed method compared to the state-of-the-art (SotA) methods.
- Convolutional Auto-encoding of Sentence Topics for Image Paragraph Generation38
A new design --- Convolutional Auto-Encoding (CAE) that purely employs convolutional and deconvolutional auto-encoding framework for topic modeling on the region-level features of an image and an architecture, namely CAE plus Long Short-Term Memory (dubbed as CAE-LSTM), that novelly integrates the learnt topics in support of paragraph generation.
- Modeling Embedding Dimension Correlations via Convolutional Neural Collaborative Filtering38
This work proposes a novel and general neural collaborative filtering framework, ConvNCF, which is featured with two designs: (1) applying outer product on user embedding and item embedding to explicitly model the pairwise correlations between embedding dimensions, and (2) employing convolutional neural network above the outer product to learn the high-order correlations amongembedding dimensions.
- 38
- 38
- MonSter++: Unified Stereo Matching, Multi-View Stereo, and Real-Time Stereo With Monodepth Priors37
A novel method that integrates monocular depth priors into multi-view depth estimation, effectively combining the complementary strengths of single-view and multi-view cues, and achieves new state-of-the-art on both stereo matching and multi-view stereo.
- On the Sample Complexity of Multichannel Frequency Estimation via Convex Optimization37
Atomic norm minimization for multichannel frequency estimation from noiseless compressive data is analyzed, showing that the sample size per channel that ensures exact estimation decreases with the increase of the number of channels under mild conditions.
- Local structure based multi-phase collaborative representation for face recognition with single sample per person37
This paper proposes local structure based multi-phase collaborative representation classification (LS_MPCRC) to solve SSPP problem, and adopts the "divide-conquer-aggregate" strategy to alleviate the dilemma of high data dimensionality and small samples.
- From text question-answering to multimedia QA on web-scale media resources37
The aim of this paper is to extend the text-based QA research to multimedia QA to tackle a range of factoid, definition and "how-to" QA in a common framework and to find multimedia answers from Web-scale media resources such as Flicker and YouTube.
- 36
- Real-Time System for Driver Fatigue Detection by RGB-D Camera36
A real-time system that utilizes RGB-D cameras to automatically detect driver fatigue and generate alerts to drivers is proposed and a novel method to predict eye states employing the WLBP feature, which is a powerful local image descriptor that is robust to noise and illumination variations.
- Concurrence-Aware Long Short-Term Sub-Memories for Person-Person Action Recognition35
A novel Concurrent Long Short-Term Memories (Co-LSTM) is proposed to model the long-term inter-related dynamics between two interacting people on the bonding boxes covering people and is shown to have superior performance compared with the state-of-the-arts methods.
- 34
- Establishment and resilience of transplanted gut microbiota in aged mice34
The results from this study systematically examining the effects of FMT on the gut homeostasis of aged hosts suggest that the compatibility between donors and recipients should be taken into account when implementing FMT.
- 34
- 34
- Learning to Rebalance Multi-Modal Optimization by Adaptively Masking Subnetworks33
This work proposes a novel importance sampling-based, element-wise joint optimization method, called Adaptively Mask Subnetworks Considering Modal Significance (AMSS), which incorporates mutual information rates to determine the modal significance and employ non-uniform adaptive sampling to select foreground subnetworks from each modality for parameter updates, thereby rebalancing multi-modal learning.
- 33
- Word2Image33
Various techniques, including the correlation analysis, semantic and visual clustering are adapted into the Word2Image system to produce sets of high quality, precise, diverse and representative images to visually translate a given word.
- IMAGHarmony: Controllable Image Editing with Consistent Object Quantity and Layout32
This work proposes IMAGHarmony, a parameter-efficient framework featuring a harmony-aware (HA) module that incorporates perception cues from the reference image into the diffusion process, and introduces a preference-guided noise selection (PNS) strategy that identifies favorable initialization conditions, substantially improving generation stability in challenging multi-object scenarios.
- 32
- Vision-based two-step brake detection method for vehicle collision avoidance32
A novel vehicle brake behavior detection method is presented by using a colorful camera or mobile device fixed on the windshield of the test car which utilized to capture the front vehicle information.
- Combining global and local matching of multiple features for precise item image retrieval32
A novel system named iSearch and global/local matching of local features are combined to do precise retrieval of item images in an interactive manner and an effective contour fragments encoding and indexing method is developed.
- 31
- 31
- 31
- 31
- 31
- Pornprobe31
PornProbe, a pornography detection system that detects pornographic contents in videos, is presented, which combines the unsupervised clustering in Latent Dirichlet Allocation and supervised learning in Support Vector Machine to achieve both high precision and recall while ensuring efficiency in both training and testing.
- Video-Text Pre-training with Learned Regions for Retrieval30
This work proposes a simple yet effective module for video-text representation learning, namely RegionLearner, which can take into account the structure of objects during pre-training on large-scale video- text pairs and is much more computationally efficient.
- 30
- 30
- Weakly-Supervised Image Hashing through Masked Visual-Semantic Graph-based Reasoning30
Experimental results on two widely-used benchmark datasets demonstrate the superiority of the proposed method for image retrieval compared with several state-of-the-art methods.
- 30
- Prototypical Prompting for Text-to-image Person Re-identification29
This paper proposes a novel prototypical prompting framework (Propot) designed to simultaneously model instance-level and identity-level matching for TIReID, and transforms the identity-level matching problem into a prototype learning problem, aiming to learn identity-enriched prototypes.
- Augmented FCN: rethinking context modeling for semantic segmentation29
The results of the joint implementation of AugNI and existing context modeling schemes show that AugFCN leads to continuous segmentation improvements in state-of-the-art context modeling.
- 29
- Crowd Counting via Multi-layer Regression29
By using multi-layer regression, MRNet achieves significant improvement in crowd counting accuracy, and outperforms the state-of-the-art methods.
- 29
- 28
- Locality-aware Cross-modal Correspondence Learning for Dense Audio-Visual Events Localization28
LoCo is presented, a Locality-aware cross-modal Correspondence learning framework for DAVE that leverages the local temporal continuity of audio-visual events as important guidance to filter irrelevant cross-modal signals and enhance cross-modal alignment throughout both unimodal and cross-modal encoding stages.
- 28
- Deep multi-person kinship matching and recognition for family photos28
The results show that the proposed Deep Kinship Matching and Recognition framework can gain superior performance on Group-Face dataset, TSKinFace dataset and FIW dataset over state-of-the-arts datasets.
- 28
- 27
- Clinical and Renal Histology Findings and Different Responses to Induction Treatment Affecting the Long-Term Renal Outcomes of Children With ANCA-Associated Vasculitis: a Single-Center Cohort Analysis27
It is demonstrated that women, MPA, and crescentic subtypes are predominant in pediatric AAV in China and initial renal failure, hypertension, sclerotic pathological subtype, and nonremission following induction treatment are predictive of long-term renal outcomes.
- Kinship-Guided Age Progression27
This work presents an efficient and effective Kinship-Guided Age Progression (KinGAP) approach for an individual, which can automatically generate personalized aging images by leveraging kinship, or more specifically, with guidance of the senior kinship face.
- 26
- 26
- Understanding the Tricks of Deep Learning in Medical Image Segmentation: Challenges and Future Directions26
This milestone work completes a comprehensive and complementary survey of the state-of-the-art MedISeg approaches and offers a practical guide for addressing the future medical image processing challenges including but not limited to small dataset learning, class imbalance learning, multi-modality learning, and domain adaptation.
- 26
- 26
- 26
- Exploring large scale data for multimedia QA26
An initial study on exploring large scale data for MMQA, which constructs a web video dataset and discusses its query strategy, statistics, feature description and groundtruth, and conducts experiments to answer definition event questions.
- 26
- STAF: 3D Human Mesh Recovery From Video With Spatio-Temporal Alignment Fusion25
A novel Spatio-Temporal Alignment Fusion (STAF) model that achieves a state-of-the-art trade-off between precision and smoothness in the recovery of 3D human mesh from monocular images.
- 25
- 25
- 24
- 24
- Age progression: Current technologies and applications24
This paper summarizes some recent and impressive age progression methods and gives some applications related to age progression, some popular face aging databases are introduced, and the future directions of the age progression are discussed.
- Towards optimizing human labeling for interactive image tagging24
This work introduces an interactive image tagging framework that can more fully make use of human's labeling efforts and proposes a dynamic multiscale clustering labeling strategy such that users can label a large group of similar regions each time.
- 24
- Surgical video workflow analysis via visual-language learning23
This work proposes a vision-language deep learning framework that incorporates intra- and inter- triplet modeling, termed I2TM, to explore the relationships among triplets and leverage the model understanding of the entire surgical process, thereby enhancing the accuracy and robustness of recognition.
- 23
- Semantic Scene Completion with Cleaner Self23
The 3D occupancy feature and the semantic relations of the “cleaner self” to supervise the counterparts of the "noisy self" to respectively address the above two incorrect predictions.
- Graph Reasoning Transformer for Image Parsing23
A novel Graph Reasoning Transformer for image parsing to enable image patches to interact following a relation reasoning pattern and results show that GReaT achieves consistent performance gains with slight computational overheads on the state-of-the-art transformer baselines.
- 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.
- 22
- Age–Period–Cohort Analysis of Trends in Mortality from Drowning in China: Data from the Global Burden of Disease Study 201522
This study demonstrated that the age-standardized mortality rates for drowning in both sexes displayed general declining trends with a decrease in the drowning mortality rate for every age group.
- Fudan-NJUST at MediaEval 2014: Violent Scenes Detection Using Deep Neural Networks.22
Results indicate that using multiple features is still very helpful, and more importantly, the proposed regularized DNN offers significantly better results than the popular SVM.
- DVF: Advancing Robust and Accurate Fine-Grained Image Retrieval with Retrieval Guidelines21
This paper designs a novel Dual Visual Filtering mechanism for the plain visual transformer, denoted as DVF, to capture subcategory-specific discrepancies and implements a discriminative model training strategy to improve the discriminability and generalization ability of DVF.
- 21
- 21
- 21
- Community question topic categorization via hierarchical kernelized classification21
A hybrid regularization approach of combining orthogonal constraint and L1 sparseness in the authors' framework to promote the discriminative power on similar topics as well as sparsing the model parameters is proposed.
- Deep supervised feature selection for social relationship recognition20
A novel Deep Supervised Feature Selection (DSFS) framework to recognize social relationships in photos is proposed, which fuses the deep learning algorithm with l2,1-norm to learn a discriminative feature subset from multi-source features by leveraging the face and body attributes.
- ViewFocus20
A novel system to explore places of interests based on the large amount of photos that are placed on Google Maps, named ViewFocus, estimates the view directions of photos via robust object matching and camera reconstruction techniques, and geo-registers the directions on the map.
- Integrated graph-based semi-supervised multiple/single instance learning framework for image annotation20
This paper proposes an integrated graph-based semi-supervised learning framework to utilize these two types of representations simultaneously, and explores an effective and computationally efficient strategy to convert the multiple-instance representation into a single-instance one.
- Look Less Think More: Rethinking Compositional Action Recognition19
A simple yet effective framework, Look Less Think More (LLTM), to reduce the strong association between visual objects and action-level labels (Look Less), and then discover the commonsense relationships between object categories and human actions (Think More).
- 19
- 19
- 19
- Low rank metric learning for social image retrieval19
A Low Rank distance Metric Learning (LRML) algorithm by discovering knowledge from these rich contextual data, to boost the performance of CBIR and formulate it as a convex optimization problem with nuclear norm minimization and propose an effective optimization algorithm based on proximal gradient method.
- Towards multi-semantic image annotation with graph regularized exclusive group lasso19
This paper proposes a general framework for harmoniously integrating the above multiple semantics, and investigates the problem of learning to annotate images with training images labeled in two or more correlated semantic spaces, such as fascinating nighttime, or exciting cat.
- ADPS: Asymmetric Distillation Postsegmentation for Image Anomaly Detection18
This work proposes an innovative approach called asymmetric distillation postsegmentation (ADPS), which employs an asymmetric distillation paradigm that takes distinct forms of the same image as the input of the teacher–student networks, driving the student network to learn discriminating representations for anomalous regions.
- 18
- 18
- 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.
- Self-iterative multiple-instance learning enables the prediction of CD4+ T cell immunogenic epitopes17
ImmuScope is introduced, a weakly supervised deep-learning framework integrating precise MHC-II antigen presentation, CD4+ T cell epitopes, and immunogenicity predictions that provides a comprehensive solution for CD4+ T cell antigen recognition and immunogenicity assessment and broad potential for advancing vaccine design and personalized immunotherapy.
- 17
- Mediapedia: Mining Web Knowledge to Construct Multimedia Encyclopedia17
A dynamic multimedia encyclopedia that takes advantage of, and in fact is built from the text and image resources on the Web, and is flexible and extensible such that it can easily incorporate new kinds of mediums such as video and languages into the framework.
- 16
- 16
- FedMLLM: Federated Fine-tuning MLLM on Multimodal Heterogeneity Data16
A benchmark to evaluate the performance of federated fine-tuning of MLLMs across various multimodal heterogeneous scenarios is introduced and extensive experimental results show that the proposed FL paradigm improves the performance of MLLMs by broadening the range of training data and mitigating multimodal heterogeneity.
- Distilling knowledge in causal inference for unbiased visual question answering16
A novel unbiased VQA model is proposed by exploring the Casual Inference with Knowledge Distillation (CIKD) to reduce the influence of bias by first constructed to explore the counterfactual causality and infer the casual target based on the causal effect.
- 16
- Unregularized Auto-Encoder with Generative Adversarial Networks for Image Generation16
A new Auto-Encoder Generative Adversarial Networks (AEGAN) is proposed, which takes advantages of both VAE and GAN and maps the random vector into the encoded latent space by adversarial training based on GAN.
- Computational face reader based on facial attribute estimation16
A novel deep convolutional network equipped with a new facial region pooling layer (called FRP-net) and the proposed fine-grained facial attribute estimation method are pushed into the face reading problem and presented to automatically infer the characteristics of a person based on his/her face.
- 16
- 15
- 15
- Deep Unpaired Blind Image Super-Resolution Using Self-supervised Learning and Exemplar Distillation15
An exemplar distillation module is developed that is able to explore useful information from exemplar HR images to constrain the deep models learned from the self-supervised learning module for the final HR image restoration.
- 15
- 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.
- 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.
- 14
- Hydration and Strength Development of Cementitious Materials Prepared with Phosphorous-Bearing Clinkers14
The results indicate that the mechanical properties of cementitious materials can be slightly improved due to the mineralization effect of the small amount of phosphorous in the clinker and significantly decreased with an increase of phosphorus.
- Image Formation Model Guided Deep Image Super-Resolution14
This work presents a simple and effective image super-resolution algorithm that imposes an image formation constraint on the deep neural networks via pixel substitution and shows that the proposed algorithm performs favorably against state-of-the-art methods.
- 14
- Robust Multiview Feature Learning for RGB-D Image Understanding14
A robust multiview feature learning approach is developed, which exploits the intrinsic relations among multiple views, leading to effective features and making the learning process robust to noises in RGB-D image understanding.
- 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.
- Multi-level semantic-aware transformer for image captioning13
This work proposes an effective method, namely Multi-Level Semantic-Aware Transformer (MLSAT) for image captioning, to simultaneously focus on contextual details and high-level semantic information centered on salient objects.
- DoFIT: Domain-aware Federated Instruction Tuning with Alleviated Catastrophic Forgetting13
Experimental results on diverse datasets consistently demonstrate that DoFIT excels in cross-domain collaborative training and exhibits significant advantages over conventional FIT methods in alleviating catastrophic forgetting.
- 13
- CAE-GReaT: Convolutional-Auxiliary Efficient Graph Reasoning Transformer for Dense Image Predictions13
This paper proposes an auxiliary and integrated network architecture, named Convolutional-Auxiliary Efficient Graph Reasoning Transformer (CAE-GReaT), which joints strengths of both CNNs and ViT into a uniform framework and proposes an efficient information diffusion strategy to reduce the computational costs in graph reasoning.
- Coupling Global Context and Local Contents for Weakly-Supervised Semantic Segmentation13
This work proposes an SS-WSSS model with only the image-level class label supervisions, termed weakly supervised feature coupling network (WS-FCN), which can capture the multiscale context formed from the adjacent feature grids, and encode the fine-grained spatial information from the low-level features into the high-level ones.
- Visual Relation of Interest Detection13
The results show that IPNet achieves the best performance compared with the baselines on visual relation detection, scene graph generation and image captioning.
- Selective Attention Network for Image Dehazing and Deraining13
To better capture various of rain and hazy details, a Selective Attention Module (SAM) is proposed to re-scale the channel-wise attention and spatial-channel attention instead of simple element-wise summation.
- 13
- Saliency-based content-aware lifestyle image mosaics13
A novel content-aware lifestyle image mosaic technique based on image saliency that creates much better mosaics in visual aspect and is compared with other mosaic methods.
- 3D Question Answering via only 2D Vision-Language Models12
It is believed that 2D LVLMs are currently the most effective alternative (of the resource-intensive 3D LVLMs) for addressing 3D tasks, while cdViews achieves state-of-the-art performance in 3D-QA while relying solely on 2D models without fine-tuning.
- 12
- 12
- Contextual and selective attention networks for image captioning12
A contextual and selective attention network that novelly memorizes contextual attention and brings out the principal components from each attention is presented, and extensive experiments on the COCO image captioning dataset demonstrate the superiority of CoSA-Net.
- Visual-Inertial State Estimation with Pre-integration Correction for Robust Mobile Augmented Reality12
A novel monocular visual-inertial state estimation approach for robust and accurate pose estimation even for low-cost IMUs and seamlessly integrate the IMU pre-integration correction module into a tightly-coupled, sliding-window based optimization framework for state estimation.
- Progressive Image Enhancement under Aesthetic Guidance12
A progressive image enhancement framework, which generates an expected "good" retouched image with a group of self-interpretable image filters under the guidance of an aesthetic assessment model, and effectively alleviates the shortage of paired training samples, and eliminates the bias caused by human subjective preferences.
- 12
- Weakly-Supervised Deep Nonnegative Low-Rank Model for Social Image Tag Refinement and Assignment12
A Weakly-supervised Deep Nonnegative Low-rank model (WDNL) is introduced to improve the quality of tags by integrating the low- rank model with deep feature learning and uncovering the intrinsic relationships between images and tags.
- One person labels one million images12
A novel framework with minimal human effort to manually annotate a large-scale image corpus and a dynamic multi-scale cluster labeling strategy is proposed to manually label the clusters of similar image regions.
- 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.
- 11
- Show, Reward, and Tell11
An attribute-based Hierarchical Generative model with Reinforcement Learning and adversarial training and considers the story generator and the reward critic as adversaries, which aims to create indistinguishable paragraphs to human-level stories.
- Photo to Family Tree: Deep Kinship Understanding for Nuclear Family Photos11
This paper extends kinship recognition from two persons to a nuclear family consisting of multiple persons and proposes a novel Deep Kinship Recognition (DKR) framework, which gains superior performance on both Group-Face dataset and TSKinFace dataset, compared with state-of-the-arts.
- Person re-identification with activity prediction based on hierarchical spatial-temporal model11
A novel person re-id framework, named Hierarchical Spatial-Temporal Model (HSTM), is proposed, which can effectively predict the person activity path and reduce the search range in the real multiple cameras surveillance system.
- Captioning Videos Using Large-Scale Image Corpus11
This paper proposes a novel video captioning approach to describe videos by leveraging freely-available image corpus with abundant literal knowledge and adopts sophisticated visual hashing techniques to efficiently index and search large-scale images for relevant captions, which is of high extensibility to evolving data and the corresponding semantics.
- Semantic-aware Hashing for Social Image Retrieval11
A novel Semantic-aware Hashing method (SaH) is proposed by discovering knowledge from these social media resources to implement approximate similarity search by exploiting heterogeneous information from the textual and visual domains.
- NIF-based seam carving for image resizing11
A novel image-resizing method which considers both texture and color information of the original image and achieves robust resizing results with reducing distortion, even when the background shows a high degree of resemblance in appearance to the contained objects.
- Projective Matrix Factorization with unified embedding for social image tagging11
This paper presents a general formulation, named Projective Matrix Factorization with unified embedding (PJMF), by which social image retagging is transformed to the nearest tag-neighbor search for each image.
- Abnormal behavior recognition system for ATM monitoring by RGB-D camera11
The experimental results show that the real-time system for ATM intelligent monitoring by using Kinect has the advantages of robustness, and provides a new mean for preventing financial crimes.
- 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.
- FoundIR-v2: Optimizing Pre-Training Data Mixtures for Image Restoration Foundation Model10
This work proposes a high-capacity diffusion-based image restoration foundation model, FoundIR-v2, which adopts a data equilibrium scheduling paradigm to dynamically optimize the proportions of mixed training datasets from different tasks, and introduces an effective Mixture-of-Experts (MoE)-driven scheduler into generative pre-training.
- EMControl: Adding Conditional Control to Text-to-Image Diffusion Models via Expectation-Maximization10
The EMControl employs the Expectation-Maximization (EM) algorithm to solve the blind inverse problem during the reverse sampling process, and ensures that samples, once consistent with the guidance, are accurately mapped back onto the noisy data manifold, adhering to the data's inherent distribution.
- Rebalanced Vision-Language Retrieval Considering Structure-Aware Distillation10
This paper designs a novel multi-granularity cross-modal matching that incorporates structure-aware distillation alongside the cross-modal matching loss, and proposes a simple yet effective method to rebalance cross-modal matching by learning structure-preserved matching representations.
- 10
- SI-Net: spatial interaction network for deepfake detection10
Spatial Interaction Network (SI-Net) is proposed, a deepfake detection method to mine potential complementary and co-occurrent features between local texture and global context concurrently, concurrently.
- 10
- Computational Face Reader10
This paper proposes a novel deep convolutional network in which a facial region pooling layer FRP layer is embedded, called FRP-net and uses the searched facial region windows locates these facial attributes instead of the commonly-used sliding windows.
- 10
- Tag ranking by propagating relevance over tag and image graphs10
This paper proposes a novel tag ranking scheme through a two-stage graph-based relevance propagation approach, which takes into count not only the relationship among tags but also the relationshipamong images.
- 10
- fMRI2GES: Co-Speech Gesture Reconstruction From fMRI Signal With Dual Brain Decoding Alignment9
This paper introduces a novel approach, fMRI2GES, that allows training of fMRI-to-gesture reconstruction networks on unpaired data using Dual Brain Decoding Alignment and shows that the proposed method can reconstruct expressive gestures directly from fMRI recordings.
- 9
- 9
- Rethinking Batch Sample Relationships for Data Representation: A Batch-Graph Transformer Based Approach9
A simple yet flexible Batch-Graph Transformer (BGFormer) for mini-batch sample representations by deeply capturing the relationships of image samples from both visual and semantic perspectives is designed.
- 9
- 9
- 9
- 9
- Hardware-Efficient Guided Image Filtering for Multi-label Problem9
A hardware-efficient Guided Filter (HGF), which solves the efficiency problem of multichannel guided image filtering and yields competent results when applying it to multi-label problems with synthesized polynomial multich channel guidance.
- 9
- 9
- Tag ranking based on salient region graph propagation9
This paper proposes a novel tag ranking approach based on the salient region that considers not only the relationship between the whole images, but also the relationships between the salient regions, and enhances the accuracy of the tag ranking.
- Augmented Image Retrieval using Multi-order Object Layout with Attributes9
A novel image search method is proposed, which allows the users to indicate any number of Regions Of Interest (ROIs) within the query as well as utilize various semantic concepts and spatial relations to search images.
- 9
- 9
- Hybrid image summarization by hypergraph partition9
A framework for hybrid image summarization is presented in which social images and corresponding textual information are taken as vertices in a hypergraph and the task of image summarizing is formulated as the problem of hypergraph partition.
- Label-to-region with continuity-biased bi-layer sparsity priors9
This work investigates how to reassign the fully annotated labels at image level to those contextually derived semantic regions, namely Label-to-Region (L2R), in a collective manner, and presents a sparse representation formulation for discovering how well an image or a semantic region can be robustly reconstructed by all the other image patches from the input image set.
- Label-specific training set construction from web resource for image annotation9
A novel semi-automatic framework to construct a more accurate and effective training set from these web media resources for each label that the authors want to learn, and Locality sensitive Hashing is applied to find the most possible region candidates of a given label efficiently.
- Looking into the world on Google Maps with view direction estimated photos9
The proposed system allows users to select the places they are interested in and then focus their exploration and walk freely through the map and view what can be seen along their eyes with the automatically returned precise set of photos of the target places/angles, by filtering out photos that are pointing to other directions.
- Image Re-Emotionalizing9
A learning framework which discovers emotion-related knowledge, such as image local appearance distributions, from a set of emotion annotated images is proposed, which can yield effective and natural effects of the image affection generation process.
- 9
- 8
- HOSIG: Full-Body Human-Object-Scene Interaction Generation with Hierarchical Scene Perception8
This work bridges the critical gap between scene-aware navigation and dexterous object manipulation, advancing the frontier of embodied interaction synthesis.
- Unleashing Network Potentials for Semantic Scene Completion8
The proposed AMMNet introduces a cross-modal modulation enabling the interdependence of gradient flows between modalities, and a customized adversarial training scheme leveraging dynamic gradient competition, providing a promising direction for improving the effectiveness and generalization of SSC methods.
- Single Image Deraining Using Residual Channel Attention Networks8
A deep residual channel attention network (DeRCAN) for deraining is proposed that is able to capture the inherent properties of the feature space and thus facilitates more accurate estimations of structures and details for image deraining.
- Foreground/Background-Masked Interaction Learning for Spatio-temporal Action Detection8
A novel Foreground/Background-masked Interaction Learning framework to learn the multi-actor features by attentively interacting with the hands-down foreground and background frames and a new Actor-guided Feature Aggregation scheme that integrates foreground- and background-interacted actor features with the learnable actor-based weights are presented.
- 8
- A feedback-retransmission based asynchronous frequency hopping MAC protocol for military aeronautical ad hoc networks8
Within a light-load network, the proposed FRMA has the ability of millisecond delay and 99% reliability as well as outperforms the non-feedback-retransmission based asynchronous frequency hopping media access control protocol.
- Linear Time Illumination Invariant Stereo Matching8
A new similarity measure that is invariant to global and local affine illumination changes and outperforms the current state of the art similarity measures in terms of accuracy on the Middlebury benchmark (with radiometric differences).
- Patient safety education among chinese medical undergraduates: An empirical study8
Short-term patient safety education cannot change medical students’ stereotyped cognition on matters related to patient safety, but it can effectively enhance their knowledge of laws and regulations of patient safety.
- A Unified Geolocation Framework for Web Videos8
This article analyzes different social, visual, and textual relationships from a real-world dataset and finds four relationships with apparent geography clues that can be used for web video geolocation.
- Strong geometrical consistency in large scale partial-duplicate image search8
A new geometry verification method named as Strong Geometry Consistency (SGC), which uses the orientation, scale and location information of the local feature points to accurately and quickly remove the false matches, and a simple scale weighting strategy, which gives feature points with larger scales greater weights.
- 8
- 7
- 7
- 7
- 7
- Multi-scale Activation, Selection, and Aggregation: Exploring Diverse Cues for Fine-Grained Bird Recognition7
A novel framework for FGBR, namely Multi-scale Diverse Cues Modeling (MDCM), which explores diverse cues at different scales across various stages of a multi-scale Vision Transformer (MS-ViT) in an ``Activation-Selection-Aggregation'' paradigm is proposed.
- Diverse Semantics-Guided Feature Alignment and Decoupling for Visible-Infrared Person Re-Identification7
A novel Diverse Semantics-guided Feature Alignment and Decoupling (DSFAD) network to align identity-relevant features from different modalities into a textual embedding space and disentangle identity-irrelevant features within each modality is proposed.
- 7
- 7
- A Competitive Method for Dog Nose-print Re-identification7
This paper presents the proposed methods for dog nose-print authentication (Re-ID) task in CVPR 2022 pet biometric challenge, and proposes an automatic offline data augmentation strategy for network optimization.
- 7
- Cauchy Matrix Factorization for Tag-Based Social Image Retrieval7
This paper investigates the applicable probability distributions of tag noise and proposes a novel Cauchy Matrix Factorization (CMF) method for tag-based image retrieval and results show the robustness of CMF to noisy tags of social images.
- Hierarchical Visual Relationship Detection7
A novel VRD task named hierarchical visual relationship detection (HVRD), which encourages predictions with abstract yet compatible relationship triplets when the confidence level of the specific image content is relatively low and can handle the inevitable ambiguity of groundtruth annotation in VRD.
- Deep Semantic Multimodal Hashing Network for Scalable Multimedia Retrieval.7
A unified deep multimodal hashing framework is proposed to learn compact and high-quality hash codes by exploiting the feature representation learning, inter-modality similarity preserving learning, semantic label preserving learning and hash functions learning with bit balanced constraint simultaneously.
- Cascaded Feature Augmentation with Diffusion for Image Retrieval7
This paper transfers diffusion process from online stage to offline stage and innovatively utilize output of diffusion to augment database features in a cascaded mode, which can eliminate iteration process at query time radically.
- Image annotation refinement via 2P-KNN based group sparse reconstruction7
A new method denoted as 2PKNN-GSR (Group Sparse Reconstruction) for image annotation and label refinement, which utilizes the group sparse reconstruction algorithm to refine the annotated label results which are obtained in the first step.
- Wheel7
Wheel first partitions the layers of a CNN into two kinds of modules: convolutional module and fully-connected module, and deploys them following the proposed hybrid parallelism, which reduces the transmitted data and fully using GPUs simultaneously.
- Latent Dirichlet Truth Discovery: Separating Trustworthy and Untrustworthy Components in Data Sources7
A latent Dirichlet truth (LDT) discovery model is proposed that defines a random field over all the possible configurations of the trustworthy degrees of sources and facts, and the most probable configuration is inferred by a maximum a posteriori criterion over the observed claims.
- 7
- Term Selection and Result Reranking for Question Retrieval by Exploiting Hierarchical Classification7
This paper proposes a hierarchical question classification method with a sparse regularization to mimc user's question labeling in cQAs based on the local context of the question for term selection and reranking results and then integrating them into the proposed general question retrieval framework.
- 7
- iSearch7
This paper proposes a novel system named iSearch and global/local matching of local features are combined to do precise retrieval of item images in an interactive manner and encode the spatial context with grid representations and a simple but effective approach using triangle relations constraints is proposed for spatial consistency filtering.
- 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.
- 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.
- 6
- Microalgae-Based 3D Bioprinting: Recent Advances, Applications and Perspectives6
This review analyzes both matrices and methods of 3D bioprinting, summarizing the advancement of microalgae-based 3D bioprinting into six main domains including living building materials, biophotovoltaics, photosynthetic biosynthesis, bioremediation, tissue engineering, and food engineering.
- 6
- 6
- 6
- 6
- 6
- Bi-branch network for dynamic scene deblurring6
The proposed network is compact and shows competitive restoration accuracy with a significant reduction in computational cost and nonlocal fusion maintains low computational cost in processing high-resolution frames and operates in a patch-based manner during inference.
- 6
- What Can We Learn about Motion Videos from Still Images?6
A framework of Still-to-Motion Adaptation (SMA) for human action recognition is proposed and the better performance of action recognition compared with the state-of-the-art methods is shown.
- 6
- 5
- PhysiInter: Integrating Physical Mapping for High-Fidelity Human Interaction Generation5
Physical mapping is introduced, integrated throughout the human interaction generation pipeline, used to project target motions into a physically valid space and adjusted to adhere to real-world physics constraints while retaining their original semantic meaning.
- A recover-then-discriminate framework for robust anomaly detection5
A novel recover-then-discriminate (ReDi) framework for AD that achieves state-of-the-art recognition accuracy and surprisingly found that the abnormal area that cannot be recognized in image-level representations can be easily recognized in the feature-level representation.
- 5
- 5
- EventCrab: Harnessing Frame and Point Synergy for Event-based Action Recognition and Beyond5
This article presents a synergy-aware framework, i.e., EventCrab, that adeptly integrates the"lighter"frame-specific networks for dense event frames with the"heavier"point-specific networks for sparse event points, balancing accuracy and efficiency.
- 5
- Slowfast Diversity-aware Prototype Learning for Egocentric Action Recognition5
A new Part-to-Prototype (P2P) scheme to learn prototypes from raw videos covering the interacting objects by refining the semantic information from part level to prototype level is presented, and a new Slow-Fast Context (SFC) mechanism is designed that explores the Up/Down augmentations for the prototype representation at the semantic level.
- 5
- 5
- Multimedia retrieval by deep hashing with multilevel similarity learning5
A novel deep multimodal hashing method, termed as Deep Hashing with Multilevel Similarity Learning (DHMSL), for learning compact and discriminative hash codes, which explores multilevel semantic similarity correlations of multimedia data.
- Partially Common-Semantic Pursuit for RGB-D Object Recognition5
A novel supervised feature learning method for RGB-D object recognition, named Partially Common-Semantic Learning (PCSL), which jointly captures the complementary and consistency semantic information from RGB and depth modalities.
- 5
- 5
- 5
- Key frame vector and its application to shot retrieval5
The proposed KFV incorporates the temporal correlation and spatial relations between frames, as well as content weighting, and employs a block-based frame comparison scheme, and uses EMD to compare shots represented by KFVs.
- 5
- 5
- Cross-modal Collaborative Representation Learning for Text-to-Image Person Retrieval4
A Virtual-text Embedding Synthesizer that generates high-quality virtual-text features for cross-modal collaboration, eliminating the need for actual texts, and an Identity-bounded Matching loss is proposed to handle different types of image-text pairs distinctly, leading to more accurate cross-modal correspondences.
- EDMG: Towards Efficient Long Dance Motion Generation with Fundamental Movements from Dance Genres4
A novel dance choreography framework designed to efficiently generate creative and long-lasting dance sequences conditioning on music and dance descriptions, which is validated through extensive experiments, demonstrating that the method can both effectively and efficiently generate long-term dances suitable for various dance genres.
- PTSR: Prefix-Target Graph-based Sequential Recommendation4
A Prefix-Target Graph-based Sequential Recommendation Approach (PTSR), which constructs a prefix-target graph (PTG) to collect observed correlations among prefixes and targets and utilizes a graph neural network to model these inherent correlations, thus improving the item representations used in the predictive model.
- 4
- FTMoMamba: Motion Generation with Frequency and Text State Space Models4
A novel diffusion-based FTMoMamba framework equipped with a Frequency State Space Model (FreqSSM) and a Text State Space Model (TextSSM) that achieves superior performance on the text-to-motion generation task, especially gaining the lowest FID on the HumanML3D dataset.
- TrGa: Reconsidering the Application of Graph Neural Networks in Two-View Correspondence Pruning4
This work uses complete Transformer structures instead of context-agnostic MLPs to capture correspondence features with global context information and stronger expression capability and introduces the Concatenation Graph Node and Global Structure block to separately capture the interaction patterns among sparse correspondence attribute features and the global structural information among them.
- Heterogeneous Learning for Scene Graph Generation4
A plug-and-play Heterogeneous Learning Branch is proposed, which enhances the independent representation capability of relation features in SGG methods and significantly improves the performance of all these methods in the common evaluation criteria for SGG task.
- Usability testing of radiotherapy systems as a medical device evaluation tool to inform hospital procurement decision-making4
The study results suggest that higher perceived and physiological workloads were experienced with the XHA600E than with the Varian Trilogy, and provide valuable evidence for hospital procurement decision-making regarding the selection of a suitable radiotherapy system for radiation therapists to use.
- Object-aware Deep Network for Commodity Image Retrieval4
A deep framework for commodity image retrieval (CMIR) from the view that they are same designed commodities is proposed, which can catch as many design details as possible by exploring object detection and ranking sensitive feature learning.
- LEGO-MM: LEarning Structured Model by Probabilistic loGic Ontology Tree for MultiMedia4
A new framework is proposed, termed LEarning Structured Model by Probabilistic loGic Ontology Tree for MultiM edia (LEGO 1 -MM), which can seamlessly integrate both the new target training examples and the existing primitive concept models to infer the more complex concept models.
- A novel video dehazing method based on temporal visual coherence4
A novel video dehazing scheme which mainly consists of atmospheric light estimation, transmission map extraction and optimization by utilizing motion estimation to improve the temporal visual coherence on the transmission map sequence is proposed.
- Query expansion enhancement by fast binary matching4
This work proposes to enhance query expansion by fast binary matching to minimize the amount of irrelevant local features introduced, and achieves promising improvement in mean Average Precision (mAP) over the state-of-the-art query expansion approaches.
- Estimating Poses of World’s Photos with Geographic Metadata4
A novel method to estimate the view directions of world's photos well is presented, and the pose referenced on Google Maps using the geographic Metadata of photos is obtained using normalized 8-point algorithm.
- A projection‐based image quality measure4
A new objective quality assessment method based on signal projection for full reference model is proposed, developed by simple parameters to achieve high efficiency and low computational complexity.
- 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
- Equal is Not Always Fair: A New Perspective on Hyperspectral Representation Non-Uniformity3
This work proposes FairHyp, a fairness-directed framework that explicitly disentangles and resolves the threefold non-uniformity of HSI representations through cooperative yet specialized modules, and redefine fairness as a structural necessity in HSI modeling and offer a new paradigm for balancing adaptability, efficiency, and fidelity in high-dimensional vision tasks.
- 3
- 3
- 3
- 3
- 3
- 3
- OmniFusion: Exemplar-Based Video Colorization Using OmniMotion and DifFusion Priors3
This paper proposes a novel video colorization method called OmniFusion, which iteratively completes the video colorization through two following steps, surpassing existing methods in terms of color fidelity and visual quality.
- 3
- 3
- LUT-GCE: Lookup Table Global Curve Estimation for Fast Low-light Image Enhancement3
This work develops a novel cubic curve formulation for light enhancement, which enables an image-adaptive and pixel-independent curve for the range adjustment of an image and proposes a global curve estimation network (GCENet), a very light network with only 25.4k parameters.
- SLLEN: Semantic-aware Low-light Image Enhancement Network3
Unlike currently available approaches, the proposed SLLEN is able to fully lever the semantic information, e.g., IEF, HSF, and SS dataset, to assist LLE, thereby leading to a more promising enhancement performance.
- 3
- 3
- Instance of Interest Detection3
A Cross-influential Network for IOID is proposed, which integrates both visual saliency and semantic context for image semantic description and outperforms the state-of-the-art baselines on this dataset.
- Bidirectional Multimodal Recurrent Neural Networks with Refined Visual Features for Image Captioning3
A bidirectional multimodal Recurrent Neural Network (m-RNN) model which considers both history and future semantic context through a Bidirectional recurrent layer is proposed.
- Designing by Training: Acceleration Neural Network for Fast High-Dimensional Convolution.3
An Acceleration Network (AccNet) is proposed which turns the work of designing new fast algorithms to training the AccNet, and is able to design acceleration algorithms for a ton of convolutions including Gaussian/non-Gaussian convolutions and produce state-of-the-art results.
- Visual understanding by mining social media: recent advances and challenges3
This work conducts a comprehensive survey of recent advances in visual understanding by mining social media in order to discuss their merits and limitations and analyzes the difficulties and challenges of visual understanding followed by several possible future research directions.
- Saliency-Based Content-Aware Image Mosaics3
The image saliency is used in the whole process of creating an image mosaic with variable size tiles, while a novel energy map is proposed by combining Neighborhood Inhomogeneity Factor and Graph-Based Visual Saliency.
- The Method for Constructing Block Sparse Measurement Matrix Based on Orthogonal Vectors3
A class of highly sparse and deterministic scrambled block measurement matrices based on orthogonal vectors (SBOV) is proposed, which could improve sensing efficiency and reduce computation complexity.
- 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.
- LeCoT: revisiting network architecture for two-view correspondence pruning2
This work designs a two-view correspondence pruning network called LeCoT, which can naturally leverage global context information at different stages and outperforms state-of-the-art methods in correspondence pruning, relative pose estimation, homography estimation, visual localization, and 3D reconstruction tasks.
- 2
- 2
- Deformation Detection Method for Substation Noise Barrier Column Based on Deep Learning and Digital Image Technology2
The results show that the structural displacement identification method based on CNN and DIC technology can accurately identify the displacement change in the structure, which greatly improves the efficiency of image displacement calculation using DIC technology.
- 2
- 2
- 2
- 2
- Dynamic in Static: Hybrid Visual Correspondence for Self-Supervised Video Object Segmentation2
HVC is presented, a unified self-supervised approach to learn visual representations of static-dynamic feature similarity, and proposes a hybrid visual correspondence loss to learn joint static and dynamic consistency representations.
- Video Colorization: A Survey2
This paper aims to review existing state-of-the-art video colorization methods from a novel perspective, and gains deeper insight into the evolution of existing methods in terms of spatial-temporal consistency.
- 2
- 2
- Recovering Overlapping Partials for Monaural Perfect Harmonic Musical Sound Separation Using Modified Common Amplitude Modulation2
The proposed Modified Common Amplitude Modulation (Modified-CAM) system aims to separate monaural perfect harmonic musical sound (except the unison), especially for the perfect octave interval, and can accurately recover both the amplitudes and the phases of theperfect octave, which can be considered as the typical but the hardest condition.
- Multimedia news QA: Extraction and visualization integration with multiple-source information2
A multimedia question answering system based on web news data is built and topic analysis is used to present concise and reliable answers including textual and visual information.
- 2
- 2
- 2
- Active Learning on Sparse Graph for Image Annotation2
This paper proposes a novel active learning approach based on sparse graph that selects the samples based on two criteria: uncertainty and representativeness, and shows that bringing the representsativeness criterion into the sample selection process can significantly improve the active learning effectiveness.
- 2
- Generating Class Test Case in Parallel on Multi-core Computers2
A method to generate the class test case in parallel with multi-threading is presented on multi-core computers and results show that the algorithm can remarkably reduce the required states and run time of the test case generation.
- 2
- DMAP: Human-Aligned Structural Document Map for Multimodal Document Understanding1
A Structured-Semantic Understanding Agent is designed to construct DMAP, which explicitly encodes both hierarchical organization and inter-element relationships within multimodal documents within multimodal documents, and a Reflective Reasoning Agent performs structure-aware and evidence-driven reasoning.
- 1
- 1
- 1
- Rebalanced Multimodal Learning with Data-aware Unimodal Sampling1
A novel MML approach called \underline{D}ata-aware \underline{U}nimodal \underline{S}ampling~(\method), which aims to dynamically alleviate the modality imbalance caused by sampling, which can be seamlessly incorporated into almost all existing multimodal learning approaches as a plugin.
- Richer Semantics, Better Alignment: Aligning Visual Features with Explicit and Enriched Semantics for Visible-Infrared Person Re-Identification1
This paper develops an Explicit Semantics-Guided Feature Alignment (ESFA) module, which supplements textual descriptions for cross-modality images and aligns image-text pairs within each modality, alleviating the distribution discrepancy of visual features.
- Diffusion-Guided Knowledge Distillation for Weakly-Supervised Low-Light Semantic Segmentation1
The proposed Diffusion-Guided Knowledge Distillation for Weakly-Supervised Low-light Semantic Segmentation (DGKD-WLSS) is a novel framework that synergistically combines Diffusion-Guided Knowledge Distillation (DGKD) with Depth-Guided Feature Fusion (DGF2).
- Multi-scale Activation, Refinement, and Aggregation: Exploring Diverse Cues for Fine-Grained Bird Recognition1
A novel framework for FGBR, namely Multi-scale Diverse Cues Modeling (MDCM), which explores diverse cues at different scales across various stages of a multi-scale Vision Transformer (MS-ViT) in an "Activation-Selection-Aggregation" paradigms.
- 1
- 1
- Diff-MM: Exploring Pre-trained Text-to-Image Generation Model for Unified Multi-modal Object Tracking1
Diff-MM leverages the UNet of pre-trained Stable Diffusion as a tracking feature extractor through the proposed parallel feature extraction pipeline, which enables pairwise image inputs for object tracking and introduces a multi-modal sub-module tuning method that learns to gain complementary information between different modalities.
- Mitigating Query Selection Bias in Referring Video Object Segmentation1
Triple Query Former (TQF) is proposed, which factorizes the referring query into three specialized components: an appearance query for static attributes, an intra-frame interaction query for spatial relations, and an inter-frame motion query for temporal association.
- 1
- Aligning Text-to-Image Diffusion Models to Human Preference by Classification1
Inspired by preference learning in large language models, ABC (Alignment by Classification), a simple yet effective framework for aligning diffusion models with human preferences is proposed, offering a scalable and robust solution for preference-based text-to-image fine-tuning.
- Embedding and Enriching Explicit Semantics for Visible-Infrared Person Re-Identification1
This paper proposes an Embedding and Enriching Explicit Semantics (EEES) framework to learn semantically rich cross-modality pedestrian representations and develops Explicit Semantics Embedding (ESE), which automatically supplements language descriptions for pedestrians and aligns image-text pairs into a common space, thereby learning visual content associated with explicit semantics.
- Continual Learning meets Multimodal Foundation Models: Fundamentals and Advances1
This workshop aims to provide a venue where academic researchers and industry practitioners can come together to discuss the principles, limitations and applications of multimodal foundation models in continual learning for multimedia applications, and promote the understanding of multimodal foundation models in continual learning.
- Diffuse large B-cell lymphoma with contemporary involvement of central and peripheral nervous system: A case report and literature review1
The patient presented to the hospital with fatigue, dyspnoea, and limb pain for over 7 months, accompanied by progressive breathlessness and unconsciousness in the last 6 days, and died of severe pneumonia.
- 1
- 1
- 1
- 1
- Fuzzy Model-Based Asynchronous Control for Markov Switching Systems with Stochastic Fading Channels1
This work investigates the asynchronous control for fuzzy Markov switching systems (MSSs) with randomly occurring fading channel by resorting to a T-S fuzzy model and the Rice fading model is proposed to capture the randomly occurring channel fading.
- 1
- Discriminative supplementary representation learning for novel-category classification1
A new Discriminative Supplementary Feature Learning (DSFL) method to expand the manually defined attributes by learning discriminative supplementary features to implement semantic attribute augmentation for novel-category image classification.
- Effective Action Detection Using Temporal Context and Posterior Probability of Length1
The posterior probability of proposal length is proposed to utilize to adjust the selection criterion of action proposals to effectively encourage the proposals with reasonable lengths and suppress the high-classification-score proposals with unreasonable lengths.
- 1
- 1
- What Shall I Look Like after N Years?1
An Auto Age Progression system, which automatically renders a series of aging faces in the future age ranges and generates an aging sequence (aging video) covering the entire life for an individual input.
- Local Similarity based Discriminant Analysis for Face Recognition1
A novel method called local similarity based linear discriminant analysis (LS_LDA) is proposed, which is motivated by the “divide-conquer” strategy, and has strong robustness to expression, illumination, occlusion and time variation.
- 1
- Cuteness Recognition and Localization in the Photos of Animals1
The model obtains promising performance in distinguishing cute images from non-cute ones and outperforms the classical models with not only better recognition accuracy, but also more intuitive localization of the cuteness in the images.
- Cross Concept Local Fisher Discriminant Analysis for Image Classification1
A new pairwise constraints method called Cross Concept Local Fisher Discriminant Analysis (C2LFDA) for image classification, which considers the cross concept images as a special case of within-class samples and models the semantic relations of images for distance metric learning under the framework of LFDA.
- 1
- Learning Semantics From Multimedia Web Resources: An Introduction to the Special Issue1
The thirteen papers in this special issue focus on effective techniques for learning semantics from multimedia Web resources.
- 1
- 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.
- Capturing a great photo via learning from community-contributed photo collections1
A novel system to help users, especially amateur photographers, to interactively learn to improve photograph skills by leveraging the available web image collections and users are able to get intuitive knowledge on the relationship between camera parameters and the resulted shooting effects.
- 1
- 1
- ACM SIGMM the first workshop on web-scale multimedia corpus (WSMC09)1
The Workshop will provide a forum to consolidate key factors related to research on very large scale multimedia dataset such as the construction of dataset, creation of ground truth, sharing and extension of such resources in terms of groundtruth, features, algorithms and tools etc.
- Proceedings of the 1st workshop on Web-scale multimedia corpus1
The Workshop will provide a forum to consolidate key factors related to research on very large scale multimedia dataset such as the construction of dataset, creation of groundtruth, sharing and extension of such resources in terms of ground truth, features, algorithms and tools etc.
- –
- –
- Wavelet-Based Adaptive Vision State Space Network for Image Harmonization–
A wavelet-based adaptive vision state space network to harmonize local regions by comprehensively exploiting long-range background context is proposed and a low-frequency harmonization block is introduced to aggregate global context and capture foreground–background interactions.
- –
- –
- –
- Conducting Conditional Diffusion by Estimating the Mean Vector of von Mises-Fisher Distribution–
This work proposes a new approach using the von Mises-Fisher distribution to model the denoised result, turning the conditional generation task into an estimation problem for vMF parameters, and optimizing the mean vector of the vMF distribution by minimizing the KL divergence between the prior and posterior distributions.
- Enhancing Text-based Image Captioning through Text-aware Self-critical Training–
The Text-aware Self-Critical Training (TSCT) method is introduced, which leverages a text-aware reward along with the REINFORCE algorithm to optimize TextCaptioning models and enhances the language model by integrating token type information of input words, which is obtained by collecting a sentence dataset based upon the original TextCaps dataset.
- –
- PGP-DiffSR: Phase-Guided Progressive Pruning for Efficient Diffusion-based Image Super-Resolution–
A lightweight diffusion method, PGP-DiffSR, is developed by removing redundant information from diffusion models under the guidance of the phase information of inputs for efficient image super-resolution and significantly reducing computational load and memory consumption.
- –
- PC-Diffusion: Aligning Diffusion Models with Human Preferences via Preference Classifier–
This work proposes a novel framework for human preference alignment in diffusion models (PC-Diffusion), using a lightweight, trainable Preference Classifier that directly models the relative preference between samples, and achieves comparable preference consistency to DPO.
- AccCtr: Accelerating Training-Free Conditional Control For Diffusion Models–
AccCtr is introduced, a method that simplifies the conditional sampling algorithm by maximizing the sum of two objectives, and identifies the most time-consuming steps and recommends retraining condition extraction network to reduce its computational cost.
- Generative Semantic Probing for Vision-Language Models via Hierarchical Feature Optimization–
Generative Semantic Probing (GSP) is introduced, a novel training-free framework that synthesizes images to probe the implicit semantic preferences of VLMs and enables unprecedented visualization of VLMs' decision boundaries.
- DISCO: DISCrete nOise for Conditional Control in Text-to-Image Diffusion Models–
DISCO guidance is introduced, which replaces the continuous conditional correction term with a finite codebook of discrete noise vectors sampled from a Gaussian prior, and is differentiability-free, and training-efficient, avoiding the gradient computation and architectural redundancy of prior methods.
- –
- Dual-view Pyramid Network for Video Frame Interpolation–
A dual-view pyramid network is built by introducing pyramidal dual-view correspondence matching to unleash the multifaceted knowledge yielded by the hierarchical views at multiple scales in a pyramid network and formulates frame refinement as progressive latent representation generations by developing flow-guided cross-scale attention for feature fusion among frames.
- –
- ExpRDiff: Short-exposure Guided Diffusion Model for Realistic Local Motion Deblurring–
A blur-aware guided image restoration method that utilizes sharp structural details from short-exposure images, facilitating accurate reconstruction of heavily blurred regions and a short-exposure guided diffusion model that explores useful features from short-exposure images and blurred regions to better constrain the diffusion process.
- –
- –
- The Solution for Temporal Action Localisation Task of Perception Test Challenge 2024–
This report presents the method for Temporal Action Localisation (TAL), which focuses on identifying and classifying actions within specific time intervals throughout a video sequence, and involves training both multimodal and unimodal models followed by combining their predictions using the Weighted Box Fusion (WBF) method.
- –
- –
- Reproducibility Companion Paper–
The file structure of the source code is explained and the details of the ViROI dataset are published, which can be used to retrain the model with custom parameters and list the parameters that can be modified for custom training and inference.
- Preliminary study on multi-class heart disease detection model of echocardiogram using deep learning–
- Reproducibility Companion Paper: Instance of Interest Detection–
The file structure of the source code and the details of the IOID dataset are explained, which can be used to retrain the model with custom parameters and a program for component analysis is provided to help other researchers to do experiments with alternative models that are not included in the authors' experiments.
- –
- –
- Attributes Consistent Faces Generation Under Arbitrary Poses–
This work aims to reduce the posterior distribute complexity given attributes by exploring face structure knowledge and imposing semantic consistency constraint in an end-to-end learning fashion by addressing the semantic and structure consistent problem in attribute-conditioned image generation.
- Mini Neural Networks for Effective and Efficient Mobile Album Organization–
An auto mobile album organization system, which can automatically classify daily photos in mobile devices into six daily categories, e.g., Baby, Food, Party, Scenery, Selfie, and Sport is presented.
- –
- Sketch-Based Image Retrieval with Multiple Binary HoG Descriptor–
A novel sketch-based image retrieval method leveraging Multiple Binary HoG (MBHoG) descriptor, which uses Ham-ming distance with two binary masks as constraint to re-trieve images and ensures time and space efficiency.
- –
- Study on power grid characteristics in summer based on Linear regression analysis–
Linear regression models established in this paper will promote electricity load growth forecast, fault repair work order review, distribution network operation weakness analysis and other work to further deepen the refinement.
- Temporal Action Localization Based on Temporal Evolution Model and Multiple Instance Learning–
A novel framework which firstly models each action clip based on its temporal evolution, and then adopts a deep multiple instance learning (MIL) network for jointly classifying action clips and refining their temporal boundaries is proposed.
- Matrix Entropy Driven Maximum Margin Feature Learning–
An in-depth investigation to the heteroscedastic problem in themaximum margin criterion is presented, and a new Maximum Margin Framework (MMF) is proposed based on the analysis to improve the traditional maximum margin criterion.
- –
- Analysis on the current performance and countermeasures on privilege management of medical qualifications in China–
The authors introduced their insights and specific recommendations for the design and implementation of privilege management of medical qualifications in China, including a platform for privilege management, a classified catalogue for medical qualifications, a procedure for the application, approval, examination and dynamic management of the qualifications.
- –
- –
- Joint depth map interpolation and segmentation with planar surface model–
This work proposes a unified variational formula for joint depth map interpolation and segmentation, which uses a multi-label representation of the depth map, where each label corresponds to a parametric representations of the planar surface on a segment.
- –
- Application of medical quality management based on system theory, cybernetics and information theory–
The hospital consultation is introduced as an application example of six platforms in the medical quality management.
- –
- Real-Time Head Pose Estimation by RGB-D Camera–
Experimental results have shown that the use of both depth data and RGB data in the method overcomes the shortcomings of single depth or RGB data, and is still robust when there is only one type of data reliable.
- –
- Clinical and Genetic Features of Multiple Familial Trichoepithelioma in 19 Chinese Families–
MFT is a rare, autosomal dominantly inherited disease with high penetrance, which characterized by numerous, skin-colored, dome-shaped and firm papules and nodules around nose.
- Introduction to the Special Section on Distance Metric Learning in Intelligent Systems–
This special Section presents high-quality articles that aim to advance the state-ofthe-art research on distance metric learning, with the focus on solving important and practical issues in deployingdistance metric learning to intelligent systems in various domains.
- Enlightenment form German Hospital Management System to China's Hospital Management–
There is great difference in hospital management system between China and Germany and through the analysis of the characteristics ofospital management system, the authors can learn advanced experiences.
- –
- Modified reduced-sum pyramid and one of its implementations–
A type of modified reduced-sum pyramid (MRSP) is proposed, in which every father- node can be got by adding a certain integer to one of its son-nodes, and the son-node can also beGot by subtracting the same integer from the father-node.
- –
- –
- –
- Image categorization based on multi-scale vocabulary–
A local feature based image categorization scheme by using Multi-Scale Vocabulary, which works by partitioning the feature space into clusters at several different levels to form multi-scale vocabulary and generate corresponding fixed-length descriptors at different scales for each image.
- –
- –
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.