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
- Thomas S. HuangSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works523 from public data
- Person Transfer GAN to Bridge Domain Gap for Person Re-identification2,033
A Person Transfer Generative Adversarial Network (PTGAN) is proposed to relieve the expensive costs of annotating new training samples and comprehensive experiments show that the domain gap could be substantially narrowed-down by the PTGAN.
- HunyuanVideo: A Systematic Framework For Large Video Generative Models1,773
HunyuanVideo is introduced, an innovative open-source video foundation model that demonstrates performance in video generation comparable to, or even surpassing, that of leading closed-source models.
- 4D Gaussian Splatting for Real-Time Dynamic Scene Rendering1,670
This work proposes 4D Gaussian Splatting (4D-GS) as a holistic representation for dynamic scenes rather than applying 3D-GS for each individual frame, and achieves real-time rendering under high resolutions, 82 FPS at an 800x800 resolution on an RTX 3090 GPU while maintaining comparable or better quality than previous state- of-the-art methods.
- Pose-Driven Deep Convolutional Model for Person Re-identification843
A Pose-driven Deep Convolutional (PDC) model is proposed to learn improved feature extraction and matching models from end to end and explicitly leverages the human part cues to alleviate the pose variations and learn robust feature representations from both the global image and different local parts.
- Pixel Difference Networks for Efficient Edge Detection541
A simple, lightweight yet effective architecture named Pixel Difference Network (PiDiNet) for efficient edge detection that adopts novel pixel difference convolutions that integrate the traditional edge detection operators into the popular convolutional operations in modern CNNs for enhanced performance on the task, which enjoys the best of both worlds.
- 474
- Modeling Point Clouds With Self-Attention and Gumbel Subset Sampling455
This work develops Point Attention Transformers (PATs), using a parameter-efficient Group Shuffle Attention (GSA) to replace the costly Multi-Head Attention, and proposes an end-to-end learnable and task-agnostic sampling operation, named Gumbel Subset Sampling (GSS), to select a representative subset of input points.
- Deep Representation Learning With Part Loss for Person Re-Identification439
Experimental results on three person ReID datasets, i.e., Market1501, CUHK03, and VIPeR, show that the proposed deep representation learning procedure named part loss network outperforms existing deep representations.
- GLAD410
A Global-Local-Alignment Descriptor (GLAD) and an efficient indexing and retrieval framework that leverages the local and global cues in human body to generate a discriminative and robust representation are proposed.
- ControlVideo: Training-free Controllable Text-to-Video Generation392
ControlVideo, adapted from ControlNet, leverages coarsely structural consistency from input motion sequences, and introduces three modules to improve video generation, which outperforms the state-of-the-arts on extensive motion-prompt pairs quantitatively and qualitatively.
- Dynamic Multiscale Graph Neural Networks for 3D Skeleton Based Human Motion Prediction374
Novel dynamic multiscale graph neural networks (DMGNN) to predict 3D skeleton-based human motions and outperforms state-of-the-art methods in both short and long-term predictions on the datasets of Human 3.6M and CMU Mocap.
- DR2-Net: Deep Residual Reconstruction Network for image compressive sensing359
A novel Deep Residual Reconstruction Network (DR2-Net) to reconstruct the image from its Compressively Sensed measurement by outperforms traditional iterative methods and recent deep learning-based methods by large margins at measurement rates 0.01, 0.1, and 0.25.
- 344
- Scalable Person Re-identification on Supervised Smoothed Manifold342
An unconventional manifold-preserving algorithm is proposed, which can make best use of supervision from training data, whose label information is given as pairwise constraints, scale up to large repositories with low on-line time complexity, and be plunged into most existing algorithms, serving as a generic postprocessing procedure to further boost the identification accuracies.
- GaussianDreamer: Fast Generation from Text to 3D Gaussians by Bridging 2D and 3D Diffusion Models339
This paper attempts to bridge the power from the two types of diffusion models via the recent explicit and efficient 3D Gaussian splatting representation via a fast 3D object gener-ation framework, named as GaussianDreamer, where the 3D diffusion model provides priors for initial-ization and the 2D diffusion model enriches the geometry and appearance.
- 317
- 308
- DisturbLabel: Regularizing CNN on the Loss Layer266
An extremely simple algorithm which randomly replaces a part of labels as incorrect values in each iteration, which prevents the network training from over-fitting by implicitly averaging over exponentially many networks which are trained with different label sets.
- 256
- 238
- 229
- Bottom-Up Temporal Action Localization with Mutual Regularization213
Two regularization terms are introduced to mutually regularize the learning procedure for temporal action localization, and the Intra-phase Consistency (IntraC) regularization is proposed to make the predictions verified inside each phase; and the Inter- phase Consistencies (InterC)Regularization is suggested to keep consistency between these phases.
- 212
- 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.
- 203
- 200
- Multi-Scale Triplet CNN for Person Re-Identification197
A multi-scale triplet convolutional neural network which captures visual appearance of a person at various scales is proposed, addressing the problem of small training set in person re-identification.
- 191
- 185
- 169
- Context-Aware Multi-View Summarization Network for Image-Text Matching165
A novel context-aware multi-view summarization network to summarize context-enhanced visual region information from multiple views and designs an adaptive gating self-attention module to extract representations of visual regions and words.
- Image Caption with Global-Local Attention158
This paper proposes a global-local attention (GLA) method by integrating local representation at object-level with global representation at image-level through attention mechanism that can pay more attention to how to predict the salient objects more precisely with high recall while keeping context information atimage-level cocurrently.
- 156
- 154
- 153
- 153
- Learning to photograph150
An intelligent photography system, which automatically and professionally generates/recommends user-favorite photo(s) from a wide view or a continuous view sequence, and can function as intelligent professionalview guider based on real-time view quality assessment and the embedded compass.
- Iterative Reorganization With Weak Spatial Constraints: Solving Arbitrary Jigsaw Puzzles for Unsupervised Representation Learning149
This paper presents a novel approach which applies to jigsaw puzzles with an arbitrary grid size and dimensionality and provides a fundamental and generalized principle, that weaker cues are easier to be learned in an unsupervised manner and also transfer better.
- 148
- 148
- Super-Bit Locality-Sensitive Hashing140
The extensive experiments on real data well validate that given the same length of binary code, SBLSH may achieve significant mean squared error reduction in estimating pairwise angular similarity, andSBLSH shows the superiority over SRP-LSH in approximate nearest neighbor (ANN) retrieval experiments.
- Creating Something From Nothing: Unsupervised Knowledge Distillation for Cross-Modal Hashing139
This paper proposes a novel approach that enables guiding a supervised method using outputs produced by an unsupervised method, and makes use of teacher-student optimization for propagating knowledge.
- 139
- Image Classification and Retrieval are ONE138
It is demonstrated that the essentials of image classification and retrieval are the same, since both tasks could be tackled by measuring the similarity between images, and a unified algorithm is proposed, which only involves manual object definition, regional description and nearest-neighbor search.
- Human Daily Action Analysis with Multi-view and Color-Depth Data136
A new descriptor of depth information for action representation is proposed, which depicts the structural relations of spatiotemporal points within action volume using the distance information in depth data.
- BridgeNet: A Continuity-Aware Probabilistic Network for Age Estimation128
The proposed BridgeNet, which aims to mine the continuous relation between age labels effectively, outperforms the state-of-the-art methods and can be jointly learned in an end-to-end way.
- Zigzag Learning for Weakly Supervised Object Detection125
This paper proposes a zigzag learning strategy to simultaneously discover reliable object instances and prevent the model from overfitting initial seeds, and introduces a novel masking regularization strategy over the high level convolutional feature maps to avoid over fitting initial samples.
- 121
- Blind image quality prediction by exploiting multi-level deep representations119
Thorough experiments have been conducted on five standard databases, which show that a significant improvement can be achieved by adopting multi-level deep representations from a very deep DNN model for learning an effective BIQA model, and consequently BLINDER considerably outperforms previous state-of-the-art BIZA methods for authentically distorted images.
- Evaluation of Salient Point Techniques115
This paper compares a wavelet-based salient point extraction algorithm with two corner detectors using the criteria: repeatability rate and information content and shows that extractinglor and texture information in the locations given by the salient points provides significantly improved results in terms of retrieval accuracy, computational complexity, and storage space of feature vectors.
- Enhancing Micro-video Understanding by Harnessing External Sounds110
A deep transfer model is developed which can jointly enhance the concept-level representation of micro-videos and the venue category prediction and alleviate the sparsity problem of unpopular categories by leveraging the external acoustic knowledge.
- 108
- Visualization and User-Modeling for Browsing Personal Photo Libraries108
A user-centric system for visualization and layout for content-based image retrieval and the ability of this framework to model or “mimic” users, by automatically generating layouts according to their preferences is demonstrated.
- 104
- 103
- 103
- 102
- Towards 3D Molecule-Text Interpretation in Language Models99
3D-MoLM enables an LM to interpret and analyze 3D molecules by equipping the LM with a 3D molecular encoder, and significantly surpasses existing baselines on downstream tasks, including molecule-text retrieval, molecule captioning, and more challenging open-text molecular QA tasks, especially focusing on 3D-dependent properties.
- Dense Temporal Convolution Network for Sign Language Translation99
A dense temporal convolution network, termed DenseTCN which captures the actions in hierarchical views and addresses the SLT problem by different views, including embedded short-term and extended longterm sequential learning.
- Feature representation for statistical-learning-based object detection: A review99
The feature sketch is drawn and new insights for feature utilization are provided in order to tackle future challenges of object detection through this review of feature representation in statistical learning based object detection.
- 97
- 96
- 95
- Co-Evolutionary Compression for Unpaired Image Translation93
A novel co-evolutionary approach for reducing their memory usage and FLOPs simultaneously and synergistically optimized for investigating the most important convolution filters iteratively is developed.
- 91
- 91
- Enhancing Person Re-identification in a Self-Trained Subspace87
A self-trained subspace learning paradigm for person re-ID is proposed that effectively utilizes both labeled and unlabeled data to learn a discriminative subspace where person images across disjoint camera views can be easily matched.
- UnrealPerson: An Adaptive Pipeline towards Costless Person Re-identification84
UnrealPerson is presented, a novel pipeline that makes full use of unreal image data to decrease the costs in both the training and deployment stages and is a system that can generate synthesized images of high-quality and from controllable distributions.
- 84
- 82
- 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.
- Online latent semantic hashing for cross-media retrieval81
This paper proposes an online optimization scheme towards the challenging task of learning hash functions efficiently on streaming data points, and the computational complexity and memory cost are much less than the size of training dataset at each round.
- 80
- 80
- 76
- Recognizing human group action by layered model with multiple cues74
A layered model to describe the discriminative characteristics at different granularities and present each layer with uniform statistical representation is proposed and can better represent group actions and improve the recognition accuracy.
- Haze removal method for natural restoration of images with sky72
A user study and objective assessment comparison with a variety of methods on long-shot haze images demonstrate that the proposed Fusion of Luminance and Dark Channel Prior method retains visual truth and removes the haze effectively.
- 72
- Regularized Diffusion Process for Visual Retrieval70
A new variant o diffusion process, which also operates on a tensor product graph is proposed, defined in three equivalent formulations (regularization framework, iterative framework and limit framework, respectively), which theoretically explain how this kind of diffusion process can better reveal the intrinsic relationship between objects.
- 67
- 67
- 66
- 66
- 65
- 63
- 62
- 62
- 62
- A Novel Multi-task Tensor Correlation Neural Network for Facial Attribute Prediction61
In this article, a novel multi-attribute tensor correlation neural network (MTCN) is used to predict face attributes and achieves the best performance compared with the latest multi- attribute recognition algorithms under the same settings.
- Mining flickr landmarks by modeling reconstruction sparsity60
The main idea is to regard the landmark mining problem as a process to find photos whose visual signatures can be reconstructed using other photos of this landmark region with a minimal coding length, and a sparse reconstruction scheme offers a general perspective to mine the representative photos.
- Classification of CT Brain Images of Head Trauma60
A method for automatic classification of computed tomography (CT) brain images of different head trauma types is presented and will be useful in building a content-based medical image retrieval system.
- Improving Image Restoration Through Removing Degradations in Textual Representations58
A new perspective for improving image restoration by removing degradation in the textual representations of a given degraded image by embedding an image-to-text mapper and text restoration module into CLIP-equipped text-to-image models to generate the guidance.
- 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.
- 57
- 57
- SORT: Second-Order Response Transform for Visual Recognition56
A novel approach named Second-Order Response Transform (SORT), which appends element-wise product transform to the linear sum of a two-branch network module, which augments the family of transform operations and increases the nonlinearity of the network.
- 56
- 55
- 55
- 53
- Hadamard Matrix Guided Online Hashing52
A novel supervised online hashing scheme termed Hadamard Matrix Guided Online Hashing (HMOH) is proposed, which is an orthogonal binary matrix built via Sylvester method, which is an orthogonal binary matrix built via Sylvester method.
- Fast Non-Local Neural Networks with Spectral Residual Learning52
Spectral residual learning is proposed, a novel network architectural design for achieving fully global receptive field and its equivalence to conducting residual learning in some spectral domain is shown and performance improvement by large margins is shown.
- 52
- 51
- InterActive: Inter-Layer Activeness Propagation51
InterActive is presented, a novel algorithm which computes the activeness of neurons and network connections, and achieves state-of-the-art classification performance on a wide range of image datasets.
- 51
- Continual Vision-Language Representation Learning with Off-Diagonal Information50
A new continual vision-language representation learning framework Mod-X is proposed: Maintain off-diagonal information-matriX, which improves the capability of the multi-modal model by maintaining the multi -modal representation space alignment on the old data domain during continuously fitting the new training data domain.
- Attribute Mix: Semantic Data Augmentation for Fine Grained Recognition50
Attribute Mix is a simple but effective data augmentation strategy that can significantly improve the recognition performance without increasing the inference budgets and consistently outperforms the state-of-the-art methods on challenging benchmarks.
- 50
- 50
- 49
- Liver tumour segmentation using contrast-enhanced multi-detector CT data: performance benchmarking of three semiautomated methods49
Compared with the reference standard, the overall performance of A1 and A2 is promising, and further development and validation is necessary before reliable tumour segmentation and volumetry can be widely used clinically.
- 49
- 48
- 47
- 47
- 47
- Image classification using spatial pyramid robust sparse coding47
A new image classification method by spatial pyramid robust sparse coding (SP-RSC), which tries to find the maximum likelihood estimation solution by alternatively optimizing over the codebook and local feature coding parameters, hence is more robust to outliers than traditional sparse coding based methods.
- Content-adaptive digital music watermarking based on music structure analysis47
A novel content-adaptive music watermarking technique with multiple bit hopping and hiding scheme with different embedding parameters is applied to vocal and instrumental frames of the music.
- Content-based image retrieval using wavelet-based salient points47
It is shown that extracting the color and texture information in the locations given by these points provides significantly improved results in terms of retrieval accuracy, computational complexity and storage space of feature vectors as compared to the global feature approaches.
- 46
- 46
- Fast and accurate near-duplicate image search with affinity propagation on the ImageWeb46
This paper constructs ImageWeb, a sparse graph consisting of all the images in the database, in which two images are connected if and only if one is ranked among the top of another’s initial search result, and uses HITS, a query-dependent algorithm to re-rank the images according to the affinity values.
- Learning image manifolds by semantic subspace projection46
This research model the images as lying on non-linear image subspaces embedded in the high-dimensional space and find that multiple subspaced may correspond to one semantic concept.
- 45
- 45
- Large-scale video copy retrieval with temporal-concentration SIFT45
A novel framework for large-scale video copy retrieval is proposed in which the processes of retrieval and validation are implemented at the feature and frame level and can yield comparable accuracy, compact storage size, and more efficient execution time, as well as adapt to various video transformations.
- Semi-automatic Segmentation of 3D Liver Tumors from CT Scans Using Voxel Classification and Propagational Learning45
A semi-automatic scheme was developed for the segmentation of 3D liver tumors from computed tomography (CT) images where a support vector machine (SVM) classifier was trained to extract tumor region from one single 2D slice in the intermediate part of a tumor by voxel classification.
- 45
- 44
- Similarity Matching in Computer Vision and Multimedia44
Comparing two images, or an image and a model, is the fundamental operation for any retrieval systems and similarity matching techniques are developed mostly for recognition of objects under several conditions of the distortion while similarity measures are used in applications like image databases.
- 43
- PL-ranking43
An efficient low-rank stochastic subgradient descent method is designed to solve the proposed optimization problem and the correlations between features and labels so that the relevance between the different modalities can be enhanced after mapping them into the common subspace.
- Search-oriented Micro-video Captioning42
A large-scale multimodal pre-training network regularized by five tasks to strengthen the downstream video representation and a flow-based diverse captioning model to generate different captions from consumers' search demand is presented.
- 42
- Scalable Object Retrieval with Compact Image Representation from Generic Object Regions42
A compact representation for scalable object retrieval from few generic object regions is proposed with a fusion of learning-based features and aggregated SIFT features and is evaluated on two public ground-truth datasets with promising results.
- Gradient-Regulated Meta-Prompt Learning for Generalizable Vision-Language Models41
A novel Gradient-RegulAted Meta-prompt learning (GRAM) framework that jointly meta-learns an efficient soft prompt initialization for better adaptation and a lightweight gradient regulating function for strong cross-domain generalizability in a meta-learning paradigm using only the unlabeled image-text pre-training data.
- 41
- 41
- 40
- 40
- 40
- EGroupNet38
A hierarchic approach referred to as EGroupNet for age prediction, which includes two main stages, feature enhancement via excavating the correlations among age-related attributes and age estimation based on different age group schemes, is proposed.
- 38
- 38
- 38
- 38
- 37
- 37
- Towards Reversal-Invariant Image Representation35
A reversal-invariant version of SIFT descriptor named Max-SIFT, a generalized RIDE algorithm which can be applied to a large family of local descriptors which reveals consistent accuracy gain on various image classification tasks, including scene understanding, fine-grained object recognition, and large-scale visual recognition.
- Semantic consistency hashing for cross-modal retrieval35
A Semantic Consistency Hashing method for cross-modal retrieval that significantly outperforms the existing schemes and an effective optimal algorithm is proposed to reduce the time complexity from traditional O ( N 2 ) or higher to O(N).
- Image classification using Harr-like transformation of local features with coding residuals35
This paper proposes a novel image classification method using Harr-like transformation of local features with additional consideration of coding residuals to reduce the information loss during the local feature encoding process.
- 34
- Super-Pixel Cloud Detection Using Hierarchical Fusion CNN34
Considering super-pixel level remote sensing images contain less semantic information compared with general object classification database, a Hierarchical Fusion CNN (HFCNN) is proposed, which takes full advantage of low-level features like color and texture information and is more applicable to cloud detection task.
- 34
- 33
- TapLab: A Fast Framework for Semantic Video Segmentation Tapping Into Compressed-Domain Knowledge33
A simple and effective framework, dubbed TapLab, to tap into resources from the compressed domain, which significantly reduces redundant computations of the state-of-the-art fast semantic image segmentation models, running 3 to 10 times faster with controllable accuracy degradation.
- Two-Stream Video Classification with Cross-Modality Attention33
A cross-modality attention operation, which can obtain information from other modality in a more effective way than two-stream, is proposed and a compatible block named CMA block is implemented, which is a wrapper of the proposed attention operation.
- Smooth Neighborhood Structure Mining on Multiple Affinity Graphs with Applications to Context-Sensitive Similarity33
A new algorithm called Smooth Neighborhood (SN) is proposed that mines the neighborhood structure to satisfy the manifold assumption and is adjusted to tackle multiple affinity graphs by imposing a weight learning paradigm.
- 33
- Harmonized Multimodal Learning with Gaussian Process Latent Variable Models32
This paper attempts to address the modality heterogeneity problem based on Gaussian process latent variable models (GPLVMs) to represent multimodal data in a common space by incorporating the harmonization mechanism into several representative GPLVM-based approaches.
- Semi-supervised feature selection analysis with structured multi-view sparse regularization32
A structured multi-view sparse regularization is constructed and a novel semi-supervise feature selection framework is proposed, namely Structured Multi-view Hessian sparse Feature Selection (SMHFS).
- 32
- 32
- 32
- Rethinking Few-Shot Class-Incremental Learning With Open-Set Hypothesis in Hyperbolic Geometry31
The configuration of FSCIL is rethink with the open-set hypothesis and a hyperbolic metric learning module is incorporated into the distillation-based framework to alleviate the overfitting issue and better handle the trade-off issue between the preservation of old knowledge and the acquisition of new knowledge.
- 31
- 31
- Video-Based Cross-Modal Recipe Retrieval31
This work innovatively proposes a new problem of video-based cross-modal recipe retrieval and thoroughly investigate this issue under the attention paradigm, and proposes a parallel-attention network to independently learn the representations of videos and recipes.
- Efficient discrete supervised hashing for large-scale cross-modal retrieval31
A supervised cross-modal hashing method based on matrix factorization dubbed Efficient Discrete Supervised Hashing (EDSH) is proposed, which produces a superior performance in both accuracy and scalability over some existing cross- modal hashing methods.
- Comprehensive Distance-Preserving Autoencoders for Cross-Modal Retrieval31
Compared with eight state-of-the-art cross-modal retrieval methods, the experimental results demonstrate that the CDPAE outperforms all the unsupervised methods and performs competitively with the supervised methods.
- 31
- Adaptive Mutual Supervision for Weakly-Supervised Temporal Action Localization30
A novel Adaptive Mutual Supervision framework with two branches, where the base branch detects the most discriminative action regions, while the supplementary branch localizes the less discriminatives through an adaptive sampler, which significantly outperforms state-of-the-art methods.
- Handwritten Chinese Font Generation with Collaborative Stroke Refinement30
The proposed model significantly outperforms the state-of-the-art methods under practical setting, i.e. with only 750 paired training samples, and is added on top of any method for font synthesis.
- Adaptively Unified Semi-supervised Learning for Cross-Modal Retrieval30
A novel dragging technique is fused into a unified linear regression model to make each sample moving to the direction of its relevant label while far away from that of its irrelevant ones, and an efficient optimization algorithm is developed to learn a discriminative common subspace for different modalities.
- 30
- Image-level classification by hierarchical structure learning with visual and semantic similarities30
A novel image classification method is proposed by automatically learning the image-level hierarchical structure (ILHS) using both visual and semantic similarities of images to generate new representations.
- 30
- Multi-modal Prompting for Low-Shot Temporal Action Localization29
This paper adopts a Transformer-based two-stage action localization architecture with class-agnostic action proposal, followed by open-vocabulary classification, and proposes to prompt the pre-trained CLIP text encoder either with detailed action descriptions, or visually-conditioned instance-specific prompt vectors.
- 29
- Learning to Select Base Classes for Few-Shot Classification29
This paper utilizes a simple yet effective measure, the Similarity Ratio, as an indicator for the generalization performance of a few-shot model, and forms the base class selection problem as a submodular optimization problem over Similarity ratio.
- Single Camera Training for Person Re-Identification29
This paper investigates ReID in an unexplored single-camera-training (SCT) setting, where each person in the training set appears in only one camera, and proposes a novel loss function named multi- camera negative loss (MCNL), a metric learning loss motivated by probability suggesting that in a multi-camera system, one image is more likely to be closer to the most similar negative sample in other cameras than to themost similarnegative sample in the same camera.
- 29
- 29
- S3MKL29
This paper proposes a scalable semi-supervised multiple kernel learning method (S3MKL) that is suitable for real world image classification and personalized web image re-ranking with very little user interaction and designs a hashing system where multiple kernel locality sensitive hashing (MKLSH) are constructed with respect to different kernels.
- Semi-Autoregressive Image Captioning28
Experimental results on the MS COCO benchmark demonstrate that the proposed Semi-Autoregressive Image Captioning (SAIC) model outperforms the preceding non-autoregressive image captioning models while obtaining a competitive inference speedup.
- 28
- Image classification by search with explicitly and implicitly semantic representations28
This paper proposes a novel image classification method by explicitly and implicitly representing the images with searching strategy and evaluates the effectiveness of the proposed EISR on the Scene-15 dataset, the MIT-Indoor datasets, the UIUC-Sports dataset and the PASCAL VOC 2007 dataset.
- 28
- Parameter Efficient Fine-Tuning via Cross Block Orchestration for Segment Anything Model27
This paper equips PEFT with a cross-block orchestration mechanism to enable the adaptation of the Segment Anything Model to various downstream scenarios, and introduces a novel inter-block communication module, which integrates a learnable relation matrix to facilitate communication among different coefficient sets of each PEFT block's parameter space.
- Data Priming Network for Automatic Check-Out27
A new data priming method is proposed to solve the domain adaptation problem of automatic Check-Out, in which distracting background is removed from the training images using the coarse-to-fine strategy and images with realistic view angles are selected by the pose pruning method.
- 27
- 27
- 27
- 27
- Object categorization in sub-semantic space27
This paper proposes a novel object categorization method by using the sub-semantic space based image representation, which is not only more semantically meaningful than exemplar based representation but also more reliable and resistant to noise than traditional semantic spacebased image representation.
- 27
- 26
- 26
- 25
- Effective Image Retrieval via Multilinear Multi-Index Fusion25
This paper proposes a new multi-index fusion scheme for image retrieval by formulating this procedure as a multilinear-based optimization problem, so that the complementary information hidden in different indexes can be explored more thoroughly.
- Adaptive Hashing With Sparse Matrix Factorization25
This article proposes a novel hashing method termed adaptively sparse matrix factorization hashing (SMFH), which exploits sparse Matrix factorization to explore the parsimonious structures of the data and adopts an orthogonal transformation to minimize the quantization loss while deriving the binary codes.
- Relevance Estimation with Multiple Information Sources on Search Engine Result Pages25
A novel framework named Joint Relevance Estimation model (JRE), which learns the visual patterns from screenshots of search results, explores the presentation structures from HTML source codes and also adopts the semantic information of textual contents is proposed.
- 25
- 25
- 24
- 24
- 24
- One-Shot Fine-Grained Instance Retrieval24
Experiments show the OSFGIR framework achieves significantly better accuracy and efficiency than existing FGVC and image retrieval methods, thus could be a better solution for large-scale fine-grained object identification.
- 24
- Spatial pooling of heterogeneous features for image applications24
A novel framework with spatial pooling of heterogeneous features for mid-level representation of images and test the superior performance of the algorithm over state-of-the-art methods on several benchmark datasets for image classification and retrieval applications.
- 24
- Omni-GAN: On the Secrets of cGANs and Beyond23
Omni-GAN is presented, a variant of cGAN that reveals the devil in designing a proper discriminator for training the model, and can extrapolate low-resolution images to arbitrary resolution, even up to ×60+ higher resolution.
- 23
- Multi-Networks Joint Learning for Large-Scale Cross-Modal Retrieval23
A novel deep framework of multi-networks joint learning for large-scale cross-modal retrieval that can simultaneously achieve specific features adapting to cross- modal task and learn a shared latent space for images and sentences is proposed.
- Boosted random contextual semantic space based representation for visual recognition23
The proposed BRCSS can alleviate the semantic gap by using semantic space based image representation using boosted random contextual semantic spaces and is able to jointly consider the visual and semantic information of images.
- Social Embedding Image Distance Learning23
The experimental results show that the proposed SIDL approach makes a marked improvement compared to the state-of-the-art image distance metrics, and an interesting observation is given to shows that the learned image distance can better reflect human cognition.
- Beyond visual features: A weak semantic image representation using exemplar classifiers for classification23
This paper proposes a simple but effective image representation for image classification, which is denoted as the responses to a set of exemplar image classifiers, which are learned using SVM algorithm to distinguish the image from others in different classes.
- 23
- A robust and accumulator-free ellipse hough transform23
Experimental results show that the RAF-EHT is more robust than the existing EHTs in detecting the partial slightly-oblique ellipses and needs only a little memory because it is accumulator-free.
- 23
- 22
- 22
- Hybrid feature-based analysis of video’s affective content using protagonist detection22
A novel emotion recognition system that can intelligently analyze and automatically recognize video emotions and facilitate the development of similar video expert systems is presented, for applications such as video recommendations, video classification and video retrieval.
- 22
- 22
- 22
- Multidimensional Scaling on Multiple Input Distance Matrices22
This paper proposes a new algorithm called Multi-View Multidimensional Scaling (MVMDS) by considering each input distance matrix as one view, which can learn the weights of views automatically by exploring the consensus information and complementary nature of views.
- Fused one-vs-all mid-level features for fine-grained visual categorization22
By learning the mid-level features based on the popular Fisher vectors and convolutional neural network, this work boosts the classification accuracy by a considerable margin and advance the state-of-the-art performance in fine-grained visual categorization.
- Pipelining Localized Semantic Features for Fine-Grained Action Recognition22
The results show that by effectively accumulating localized semantic information into the action representation and classification pipeline, the results significantly improve the fine-grained action classification performance over the existing methods.
- Mutual Information-Based Emotion Recognition22
A general scheme for video emotion identification using mutual information-based feature selection followed by regression to measure video affective content in dimensional arousal-valence space is proposed.
- A semi-automated method for liver tumor segmentation based on 2D region growing with22
A semi-automatic method based on 2D region growing with knowledge-based constraints is proposed to segment lesions from constituent 2D slices obtained from 3D CT images, which resulted in an average score of 64.
- Automatic summarization of music videos22
Evaluating the quality of music summaries and effectiveness of the proposed summarization approach indicate that summaries generated using the proposed method are effective in helping realize users' expectations
- 21
- Discrete Robust Supervised Hashing for Cross-Modal Retrieval21
A discrete robust supervised hashing algorithm (DRSH) is proposed that is evaluated on three real-world datasets, and the results demonstrate the superiority of DRSH over several existing hashing methods.
- 21
- Correlated attribute transfer with multi-task graph-guided fusion21
The novelty of CAT-MtG2F is to encourage highly correlated attributes to share a common set of relevant low-level features and transfer the learned common structure from the source image set to the target image set.
- Learning to judge image search results21
This paper proposes a novel method to automatically identify the best search result list from a number of candidates, and proposes a preference learning model to quantitatively study the best image search result identification problem.
- Industrial Internet Robot Collaboration System and Edge Computing Optimization20
This paper models a differential-drive robot with nonholonomic constraints, then decomposes motion into obstacle avoidance, target turning, and target approaching behaviors to parameterize the control variables and improves robustness of global navigation in practice.
- 20
- Features-Enhanced Multi-Attribute Estimation with Convolutional Tensor Correlation Fusion Network20
Experimental results on benchmarks with multiple face attributes (MORPH-II, Adience Benchmark datasets, LAP-2016, and CelebA) show that the proposed approach has superior performance compared to state of the art.
- 20
- 20
- Deep Supervised Quantization by Self-Organizing Map20
The method integrates the Convolutional Neural Networks and Self-Organizing Map into a unified deep architecture that can be directly applied to classification task and visualization with little modification, and promising performances are demonstrated on these tasks in the experiments.
- 20
- Music video affective understanding using feature importance analysis20
Experimental results demonstrate the proposed novel feature importance analysis approach to select most representative arousal and valence features for arousal andValence modeling has a considerable improvement on music video affective understanding.
- 20
- Visual Recognition by Request19
A new paradigm named visual recognition by request is established, which allows for learning complicated whole-part hierarchies from highly incomplete annotations and inserting new concepts with minimal efforts, and establishes a solid baseline by integrating language-driven recognition into recent semantic and instance segmentation methods.
- Exploring rich intermediate representations for reconstructing 3D shapes from 2D images19
This work proposes to explore rich intermediate representation for 3D shape reconstruction by using a newly designed network architecture that captures the global spatial structure and the visible surface geometric structure, which are important for reconstructing high-quality 3D shapes.
- 19
- Nearest-neighbor method using multiple neighborhood similarities for social media data mining19
A novel neighborhood similarity measure is proposed which encodes both the local density information and semantic information, thus it has better generalization power than the original image-to-image similarity and enhances the scalability and robustness on diversified genres of images.
- Correlation-Based Feature Selection and Regression19
Significant reductions in terms of both mean absolute error and variation of absolute error compared with the state-of-the-art methods clearly demonstrate the effectiveness of the proposed general scheme for music video affective estimation using correlation-based feature selection followed by regression.
- 19
- Display Optimization For Image Browsing19
A novel technique for solving the problem of overlapping (obscured) images shown in the proposed 2-D display is developed and significantly improved visualization of large image sets is achieved when the proposed technique is applied.
- 18
- E2BoWs: An end-to-end Bag-of-Words model via deep convolutional neural network for image retrieval18
This work proposes an End-to-End BoWs model based on Deep Convolutional Neural Network that takes an image as input, then identifies and separates semantic objects in it, and finally outputs visual words with high semantic discriminative power.
- Incorporating visual adjectives for image classification18
This paper proposes a simple framework to integrate various types of adjectives, i.e., color, shape and location, for more powerful image representation, and shows that this approach achieves superior classification accuracy with reasonable computational overheads.
- 17
- Dilated Context Integrated Network with Cross-Modal Consensus for Temporal Emotion Localization in Videos17
A cross-modal consensus learning paradigm is introduced, which leverages the inherent semantic consensus between the aligned video and subtitle to achieve weakly-supervised learning on temporal emotion localization.
- Fast discrete cross-modal hashing with semantic consistency17
A Fast Discrete Cross-modal Hashing method that firstly leverages both class labels and the pair-wise similarity matrix to learn a sharing Hamming space where the semantic consistency can be better preserved, and proposes an asymmetric hash codes learning model to avoid the challenging issue of symmetric matrix factorization.
- 17
- Self-supervised learning based on discriminative nonlinear features for image classification17
KDEM is extended to a nonlinear kernel algorithm, Kernel DEM (KDEM), and evaluates KDEM extensively on benchmark image databases and synthetic data, showing the effectiveness of the approach.
- Cross-modal Self-Supervised Learning for Lip Reading: When Contrastive Learning meets Adversarial Training16
This work proposes a novel self-supervised learning framework called Adversarial Dual-Contrast Self-Supervised Learning (ADC-SSL), to go beyond previous methods by explicitly forcing the visual representations disentangled from speech-unrelated information.
- 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.
- 16
- 16
- Visual word expansion and BSIFT verification for large-scale image search16
This paper proposes constructing a supporting visual word table for all visual words by visual word expansion and presents a matching verification scheme based on binary SIFT (BSIFT) signature, which greatly improves the precision of large-scale image search.
- Mining spatiotemporal video patterns towards robust action retrieval16
This paper introduces an attention shift scheme to detect and partition the focused human actions from YouTube videos, and leverages a boosting based feature selection to output the final action descriptors, which incorporates the ranking distortion of the conjunctive queries into the boosting objective.
- Universal-to-Specific Framework for Complex Action Recognition15
An effective universal-to-specific (U2S) framework for complex action recognition is proposed, and the relationship between different classes is visualize, showing that U2S indeed improves the discriminability of learned features.
- Network Adjustment: Channel Search Guided by FLOPs Utilization Ratio15
A new framework named network adjustment is presented, which considers network accuracy as a function ofFLOPs, so that under each network configuration, one can estimate the FLOPs utilization ratio (FUR) for each layer and use it to determine whether to increase or decrease the number of channels on the layer.
- Hierarchical deep semantic representation for visual categorization15
This paper proposes Hierarchical Deep Semantic Representation (H-DSR), a hierarchical framework which combines semantic context modeling with visual features and verifies the effectiveness of H- DSR for visual categorization on two publicly available datasets.
- Interpretation of users’ feedback via swarmed particles for content-based image retrieval15
An integrated users-feedback and learning algorithm by screening individual elements of content features and driving a group of swarmed particles inside the feature space to provide a possible solution for content-based image retrieval applications is proposed.
- Image classification using boosted local features with random orientation and location selection15
A novel boosted local features method with random orientation and location selection is proposed to combine the local feature extraction and encoding with classifier training into a unified framework and gradually improve the image classification performance.
- 14
- 14
- Interactive social group recommendation for Flickr photos14
An interactive group recommendation framework with Human-in-the-Loop that is able to provide accurate group recommendations with quite a small amount of user efforts is proposed.
- Image Classification Using Spatial Pyramid Coding and Visual Word Reweighting14
An improved BoW representation using spatial pyramid coding (SPC) and visual word reweighting is presented and a relaxed but simple solution for semantic embedding into visual words is proposed.
- 14
- 14
- An algorithm to estimate mean vehicle speed from MPEG Skycam video14
This is the very first work in which stationary motion (speed) of moving objects can be estimated with high accuracy directly from MPEG motion vectors and it can be applied to other applications where the stationary motion assumption is satisfied.
- 13
- User Behavior Simulation for Search Result Re-ranking13
Two different simulation environments for the offline training of the RL ranking agent are proposed: the Context-aware Click Simulator (CCS) and the Fine-grained User Behavior Simulator with GAN (UserGAN) and a User Behavior Simulation for Reinforcement Learning (UBS4RL) re-ranking framework is designed.
- Attacking Image Captioning Towards Accuracy-Preserving Target Words Removal13
A novel adversarial learning approach is proposed for this new task, termed Show, Mask, and Tell (SMT), which crafts adversarial examples to mask the target concepts via minimizing an objective loss while training the noise generator.
- Cloud Detection Using Super Pixel Classification and Semantic Segmentation13
This paper introduces a cloud detection method based on super pixel level classification and semantic segmentation and shows that compared with other cloud detection methods, the proposed method achieves better performance.
- A multiple kernel density clustering algorithm for incomplete datasets in bioinformatics13
Compared with existing density clustering algorithms and parameter-free clustering algorithm, the proposed MKDCI algorithm tends to automatically produce clusters of better quality on the incomplete dataset in bioinformatics.
- Linear Distance Preserving Pseudo-Supervised and Unsupervised Hashing13
A novel and general hashing framework, which simultaneously considers a new linear pair-wise distance preserving objective and point-wise constraint, and a pseudo-supervised and unsupervised hashing method, in which quantization loss is considered.
- Joint image representation and classification in random semantic spaces13
A novel algorithm is proposed by combining image representation and classification in the random semantic spaces by combining local features with the sparse coding technique and using the encoding parameters for raw image representation to achieve better performances.
- Heterogeneous Semantic Level Features Fusion for Action Recognition13
This paper aims to apply the success of static image semantic recognition to the video domain, by leveraging both static and motion based descriptors in different stages of the semantic ladder, by employing a scalable method to fuse these features.
- 13
- 13
- Search by Detection: Object-Level Feature for Image Retrieval13
A novel image representation for image retrieval is proposed that identifies some regions of interest with an advanced general object detector, and the regions are described by features extracted with convolutional neural network (CNN).
- Nearest-neighbor classification using unlabeled data for real world image application13
A novel Nearest-Neighbor method, which improves the original NN approaches from three aspects and proposes a novel neighborhood similarity measure, where the similarity between test images and labeled images in the database is calculated jointly by the original image-to-image similarity and the average similarity of their neighboring unlabeled data.
- Personalized MTV Affective Analysis Using User Profile13
A novel approach combining support vector regression and psychological affective model to achieve personalized affective analysis is proposed and the experimental results illustrate the effectiveness and advantages of the proposed method.
- 13
- 12
- 12
- 12
- Improving context-sensitive similarity via smooth neighborhood for object retrieval12
A new algorithm called Smooth Neighborhood (SN) is proposed that mines the neighborhood structure to satisfy the manifold assumption and can yield state-of-the-art performances on shape retrieval, image retrieval and 3D model retrieval.
- Pseudo-positive regularization for deep person re-identification12
A Pseudo-Positive Regularization method to enrich the diversity of the training data and reduce the risk of over-fitting when training the convolutional neural network (CNN) in person re-ID.
- Neighborhood geometry based feature matching for geostationary satellite remote sensing image12
Experimental results on 25 pairs of newly annotated images show that the proposed method is competitive to several prior arts with respect to matching accuracy, and significantly more efficient than others.
- Augmented Feature Fusion for Image Retrieval System12
This work adopts the reciprocal neighbor based graph fusion approach for feature fusion and explicitly augments the original approach with the following two strategies: 1) investigates the most suitable feature combinations on various datasets, including the deep learning feature, which has been popular for image retrieval recently; 2) further improves the robustness of originalgraph fusion approach by the SVM prediction strategy.
- Superimage: Packing Semantic-Relevant Images for Indexing and Retrieval12
This work proposes to package semantically relevant images into superimages, and then index superimages instead of single images, which effectively packages multiple images into one new unit, hence significantly decreases the number of images to be indexed.
- Interactive Boosting for Image Classification12
The proposed boosting scheme merges adaboost, feature re-weighting and relevance feedback into one framework and exploits the favorable attributes of these methods and achieves superior performance over AdaBoost and other state-of-the-art projection-based classifiers.
- 12
- 12
- MultiPRE12
Experiments show that the proposed framework achieved significantly better performance in terms of both the mean and the variance of average precision than the best run reported in ImageCLEF2004.
- O-DisCo-Edit: Object Distortion Control for Unified Realistic Video Editing11
O-DisCo-Edit is proposed, a unified framework that incorporates a novel object distortion control (O-DisCo) that flexibly encapsulates a wide range of editing cues within a single representation and enables efficient, high-fidelity editing through an effective training paradigm.
- OVMR: Open-Vocabulary Recognition with Multi-Modal References11
This method, named OVMR, adopts two innovative components to pursue a more robust category cues embedding, and is a plug-and-play module that works well with exemplar images randomly crawled from the Internet.
- 11
- Birds of a feather flock together: Visual representation with scale and class consistency11
This paper proposes a scale and class consistent local-feature encoding method for image representation, which is achieved through the dense extraction of local features in different scale spaces, and the subsequent learning of the encoding parameters.
- Geometric Neural Phrase Pooling: Modeling the Spatial Co-occurrence of Neurons11
The Geometric Neural Phrase Pooling algorithm is proposed, which acts as a new type of hidden layer, which punishes the isolated neuron responses after convolution, and can be inserted into a CNN model with little extra computational overhead.
- 11
- Cluster-sensitive Structured Correlation Analysis for Web cross-modal retrieval11
A set of cluster-sensitive correlation sub-models are learned instead of a unified correlation model, which better fits the content divergence in different modalities, and achieves better model generality and bias-variance trade-off.
- Binary SIFT11
This paper proposes to generate binary SIFT from the original SIFT descriptor, and demonstrates the effectiveness and efficiency of the proposed approach by conducting partial-duplicate image search on a one-million image database.
- 11
- 11
- Boosting Multiple Classifiers Constructed by Hybrid Discriminant Analysis11
A set of hybrid dimension reduction schemes is constructed by unifying principal component analysis (PCA) and linear discriminant analysis (LDA) in a single framework and boosts both the individual features as well as a set of weak classifiers.
- 10
- 10
- UHCL-Darknet10
UHCL-Darknet is a scalable and portable DNN framework for heterogeneous clusters, and achieves 1.9x and 2.2x speedups on average respectively for the image throughput of data-parallel training and inference on the experimental heterogeneous cluster.
- 10
- Laplacian affine sparse coding with tilt and orientation consistency for image classification10
A Laplacian affine sparse coding algorithm which combines the tilt and orientations of affine local features as well as the dependency among local features is proposed, and tilt and orientation smooth constraints are added into the objective function of sparse coding.
- TBIdoc: 3D content-based CT image retrieval system for traumatic brain injury10
The TBIdoc system is presented: a content-based image retrieval (CBIR) system which works on the TBI CT images, which is expected to improve the current hospital data management in TBI and to give better support for the clinical decision-making process.
- 10
- Nonparametric motion model with applications to camera motion pattern classification10
The mean shift procedure is employed to propose a novel nonparametric motion representation that can be applied to camera motion pattern classification on 23191 MVF extracted from MPEG-7 dataset.
- Nonparametric color characterization using mean shift10
Experimental results suggest that the color characterization model is much more effective for video indexing and browsing, particularly in the domain of structured video (e.g. sports video).
- Visualization, Estimation and User-Modeling for Interactive Browsing of Image Libraries10
A user-centric system for visualization and layout for content-based image retrieval and browsing and a novel subspace feature weighting technique is proposed and used to modify 2-D layouts in a variety of context-dependent ways.
- Visualization and Layout for Personal Photo Libraries10
The experimental results show a more perceptually intuitive and informative visualization of traditional CBIR-based retrievals, providing not only a better understanding of the query context but also aiding the user in forming new queries.
- 9
- 9
- A novel two-stream saliency image fusion CNN architecture for person re-identification9
A saliency image embedding as a pedestrian descriptor is proposed and it is shown that the learned pedestrian descriptor by the proposed SIF CNN architecture provides a significant improvement over the baselines and produces a competitive performance compared with the state-of-the-art person re-ID methods on three large-scale people re- ID benchmarks.
- Supervised Coarse-to-Fine Semantic Hashing for cross-media retrieval9
A supervised hashing method for cross-media retrieval which utilizes the coarse-to-fine semantic similarity to learn a sharing space and an orthogonal rotation matrix is learned by minimizing the quantization loss while preserving the optimality of the relaxed solution.
- A two-step approach to describing web topics via probable keywords and prototype images from background-removed similarities9
By removing background similarities, a devcovolution model is utilized to strip the similarities among non-informative words/images during web topic detection, and the background-removed similarities are reconstructed to identify the probable keywords and prototype images during topic description.
- 9
- Multi-order visual phrase for scalable image search9
Multi-order visual phrase is proposed which contains two complementary clues: center visual word quantized from the local descriptor of each image keypoint and the visual and spatial clues of multiple nearby keypoints and is more robust to quantization error than existing visual phrases.
- Spatial pooling for transformation invariant image representation9
This work proposes novel spatial pooling layouts to address various transformations, and generates a more general image representation that is compact and consistent with the BoW model, which makes it applicable to image retrieval task as well.
- 9
- Visual ContextRank for web image re-ranking9
Evaluation on object images collected from the Google search engine shows that the proposed ContextRank approach outperforms VisualRank, and is comparable with the state-of-art, which is suitable for potential requirement of large scale web image retrieval.
- 9
- Report for Annotation task in ImageCLEFmed 20059
Experimental results show that using a combination of the blob region feature and three low resolution pixel maps can achieve the highest recognition accuracy.
- ChatModel: Automating Reference Model Design and Verification with LLMs8
ChatModel, an LLM-aided agile reference model generation and verification platform that streamlines the transition from design specifications to fully functional reference models by integrating design standardization and hierarchical agile modeling, is introduced.
- Global or Local: Constructing Personalized Click Models for Web Search8
This work proposes a Click Model Personalization framework (CMP) that adaptively selects from global and local models for individual users and reveals that capturing personalized behavior patterns is more important than modeling personalized relevance assessments in constructing personalized click models.
- 3D sketching for 3D object retrieval8
3D sketch and model matching problems are investigated, a novel 3D sketch-based model retrieval algorithm CNN-SBR based on Convolutional Neural Networks is proposed and the best performance in the SHREC track is achieved.
- Search Result Reranking with Visual and Structure Information Sources8
Experimental results show that the proposed two models achieve better performance than state-of-the-art ranking solutions as well as the original rankings of commercial search engines.
- 8
- Copy-Move Forgery Localization Using Convolutional Neural Networks and CFA Features8
A convolutional neural network (CNN)-based method to accurately localize the tampered regions, which combines color filter array (CFA) features, which introduces the correlation and consistency among the pixels, which can be easily destroyed by most image processing operations.
- Guest Editorial: Large-Scale Multimedia Data Retrieval, Classification, and Understanding8
The papers in this special section focus on multimedia data retrieval and classification via large-scale systems, with a view to exploiting the commonalities and differences between different tasks, e.g., image retrieval and Classification have much in common while different indexing methods evolve in a mutually supporting way.
- 8
- Learning attribute-aware dictionary for image classification and search8
An Attribute aware Dictionary Learning (AttrDL) scheme to learn multiple sub-dictionaries with specific semantic meanings and shows that the proposed scheme outperforms the state-of-the-art algorithms in both image classification and search tasks.
- Visual query attributes suggestion8
A novel query suggestion scheme named Visual Query Attributes Suggestion (VQAS) for image search with QBE, where informative attributes are suggested to the user as complements to the query.
- Multimodal LLM-Guided Semantic Correction in Text-to-Image Diffusion7
MLLM Semantic-Corrected Ping-Pong-Ahead Diffusion (PPAD), a novel framework that, for the first time, introduces a Multimodal Large Language Model as a semantic observer during inference, offering strong generality and scalability.
- 7
- 7
- Discovering Latent Topics by Gaussian Latent Dirichlet Allocation and Spectral Clustering7
This work applies Gaussian latent Dirichlet allocation (GLDA) on the diversity issue of image retrieval, and proposes a two-phase spectral clustering strategy, called dual spectral clustersering, to generate clusters from region level to image level.
- 7
- Hyperspectral image secure retrieval based on encrypted deep spectral–spatial features7
Experimental results show that the proposed method for hyperspectral image secure retrieval based on encrypted deep spectral–spatial features can effectively protect the security of image content with sufficient image retrieval accuracy.
- Structured Stochastic Recurrent Network for Linguistic Video Prediction7
This work introduces a new task of Linguistic Video Prediction (LVP), which aims to predict the forthcoming events based on past video content and generate corresponding linguistic descriptions, and proposes an end-to-end probabilistic approach named structured stochastic recurrent network (SRN) to characterize the one- to-many connections between past visual clues and possible future events.
- 7
- Automatic Endmember Extraction Using Pixel Purity Index for Hyperspectral Imagery7
Compared with traditional APPI algorithm, the experimental results show that the proposed algorithm can obtain more endmembers as well as improve the accuracy of endmembers.
- 7
- Scalable mobile search with binary phrase7
This paper proposes to build binary phrase which employs fast binary local descriptor and extracts spatial clues to improve both the efficiency and discriminative power of mobile search systems.
- Learning heterogeneous data for hierarchical web video classification7
An AdaBoost-Like transfer learning algorithm is proposed to transfer the knowledge learned from Flickr images to YouTube video domain and thus it facilitates video classification, and a bottom-up semantic forest of video concepts by learning from meta-data is constructed.
- Semi-automatic Flickr Group Suggestion7
A semiautomatic group suggestion approach with Human-in-the-Loop is proposed, which is able to provide accurate group suggestions with quite a small amount of user effort.
- 7
- 7
- 7
- A novel content-based medical image retrieval method based on query topic dependent image features (QTDIF)7
The experimental results show that the proposed QTDIF-based CBIR can provide significantly better performance than systems based general features only.
- 6
- Multi-branch Body Region Alignment Network for Person Re-identification6
A Multi-branch Body Region Alignment Network (MBRAN), to learn discriminative representations for person Re-ID, and extensive experiments have indicated that this method outperforms several state-of-the-art methods on two mainstream person re-identification datasets.
- Deep Scalable Supervised Quantization by Self-Organizing Map6
The method integrates the Convolutional Neural Networks and Self-Organizing Map into a unified deep architecture and optimizes supervised quantization loss as well as classification loss, and can be directly applied to visualization with little modification.
- Multiscale target extraction using a spectral saliency map for a hyperspectral image6
A multiscale target extraction method using a spectral saliency map is proposed for a hyperspectral image and the experimental results show that the proposed method can effectively improve the accuracy of target extraction for hyperspectrals.
- 6
- A compact method for prostate zonal segmentation on multiparametric MRIs6
A novel compact method to segment the prostate and its zones using multi-parametric magnetic resonance imaging (MRI) and the anatomical priors is presented and has the potential to be used in clinical practices.
- Undo the codebook bias by linear transformation for visual applications6
An alternative optimization algorithm is proposed to jointly search for the optimal linear transformation matrixes and the encoding parameters and show the effectiveness of the proposed method on image classification experimental results.
- Segmentation of Skull Base Tumors from MRI Using a Hybrid Support Vector Machine-Based Method6
A novel three-stage hybrid SVM (HSVM) is proposed and applied for the segmentation of skull base tumor and Quantitative results suggested that the developed method achieved significantly higher segmentation accuracy than OSVC and BSVC.
- Efficient Advertisement Discovery for Audio Podcast Content Using Candidate Segmentation6
This paper proposes a novel efficient advertisement discovery approach for large audio podcasting collections that offers a significant improvement on search speed with sufficient accuracy and has been tested over a variety of podcast contents collected from MIT Technology Review, Scientific American, and Singapore Podcast websites.
- 6
- 6
- 6
- 6
- 5
- 5
- Real Time Sales Forecasting in Omnichannel Retail Using a Hadoop Based Hybrid CNN–LSTM Deep Learning Framework5
This study proposes a Hadoop based hybrid CNN–LSTM deep learning framework that leverages Hadoop’s distributed computing capabilities to process almost 20 million multi source sales records collected over two years, satisfying second level business decision requirements.
- Text-Animator: Controllable Visual Text Video Generation5
Quantitative and qualitative experimental results demonstrate the superiority of the proposed Text-Animator approach to the accuracy of generated visual text over state-of-the-art video generation methods.
- Seek for Incantations: Towards Accurate Text-to-Image Diffusion Synthesis through Prompt Engineering5
This paper proposes a framework to learn the proper textual descriptions for diffusion models through prompt learning by utilizing the quality guidance and the semantic guidance derived from the pre-trained diffusion model to improve the matches between the input text and the generated images.
- Lformer: Text-to-Image Generation with L-shape Block Parallel Decoding5
Lformer, a semi-autoregressive text-to-image generation model that achieves faster speed than the existing transformer-based methods while keeping good generation quality and can edit images without the requirement for finetuning.
- Prototype-guided Cross-task Knowledge Distillation for Large-scale Models5
A Prototype-guided Cross-task Knowledge Distillation (ProC-KD) approach to transfer the intrinsic local-level object knowledge of a large-scale teacher network to various task scenarios and proposes a prototype learning module to learn from the essential feature representation of objects in the teacher model.
- 5
- 5
- 5
- Region similarity arrangement for large-scale image retrieval5
A promising method of geometric verification to improve the precision of Bag-of-Words (BoW) model in image retrieval by exploiting the spatial arrangement of interest regions, named Region Similarity Arrangement (RSA).
- Socio-mobile landmark recognition using local features with adaptive region selection5
This work proposes a novel socio-mobile visual recognition method using local features with adaptive region selection based local features for sodo-mobile landmark recognition that achieves the state-of-the-art performance on the Stanford mobile visual search dataset and the San Francisco landmark dataset.
- Stereotime5
This work presents a wireless 2D and 3D switchable video communication to handle the previous challenges, and name it as Stereotime, and shows the functionalities and compatibilities on 3D mobile devices in WiFi network environment.
- Distribution-Aware Locality Sensitive Hashing5
A new LSH scheme, called Distribution-Aware LSH (DALSH), is proposed to address the problem of lacking adaptation to real data, which is the intrinsic limitation of most LSH methods.
- Visual block link analysis for image re-ranking5
A Visual Block Rank approach for image re-ranking that considers images as visual block sets and analyzes the authority of images by exploring the underlying visual block link structure via LSA (Latent Semantic Analysis).
- Adaptive discriminant analysis for microarray-based classification5
Adaptive Discriminant Analysis (ADA) is proposed, which effectively exploits favorable attributes of both BDA and LDA and avoids their unfavorable ones and can find a good discriminative subspace with adaptation to different sample distributions.
- 5
- 5
- Error metric analysis and its applications5
By designing error metrics, a much richer set of distance measures besides the conventional Euclidean distance or SSD and the Manhattan distance or SAD are obtained and the corresponding nonlinear estimations such as harmonic mean, geometric mean, as well as their generalized nonlinear operations are derived.
- 5
- PDH: a human-centric interface for image libraries5
A novel optimized layout technique for large image sets, which respects (context-sensitive) mutual similarities as visualized on a shared 2-D display (a table-top) and shows a more perceptually intuitive and informative visualization of traditional CBIR-based retrievals.
- From Pattern Recognizers to Personalized Companions: A Survey of Large Language Models in Mental Health4
Viewing the field through this developmental lens, this survey provides a comprehensive synthesis of existing work, an insightful narrative of its trajectory, and a clear roadmap for future innovation in responsible, effective, and human-centered AI for mental healthcare.
- STDE-YOLOv5s: a traffic sign detection algorithm based on context information enhancement4
YOLOv5s is selected as the base model and an improved traffic sign detection algorithm is proposed, and a Context Information Enhanced Module that combines dilated convolution and deformable convolution is proposed, which enhances the network’s ability to extract context information.
- OpenDance: Multimodal Controllable 3D Dance Generation with Large-scale Internet Data4
OpenDance5D is built, an extensive human dance dataset comprising over 101 hours across 14 distinct genres and has five modalities to facilitate robust cross-modal learning.
- VioLET: Vision-Language Efficient Tuning with Collaborative Multi-modal Gradients4
A novel PET approach called VioLET (Vision Language Efficient Tuning) that utilizes collaborative multi-modal gradients to unlock the full potential of both modalities and consistently outperforms several state-of-the-art methods on 11 datasets, showcasing its superiority over existing PET approaches.
- 4
- 4
- A half-precision compressive sensing framework for end-to-end person re-identification4
Compressive sensing approaches are useful for end-to-end person re-identification (Re-ID) in reducing the overheads of transmitting and storing video frames in distributed multi-camera systems, but the reconstruction quality degrades appreciably as the measurement rate decreases for existing CS methods.
- Scalable Bag of Selected Deep Features for Visual Instance Retrieval4
A novel Bag- of-Deep-Visual-Words (BoDVW) model for instance retrieval that achieves respectable performance in comparison to other state-of-the-art methods and is proved to be more effective and efficient on large scale datasets.
- 4
- Deep hashing with top similarity preserving for image retrieval4
A new Deep Hashing with Top Similarity Preserving (DHTSP) method is proposed to optimize the quality of hash codes for image retrieval by utilizing AlexNet to extract discriminative image representations directly from the raw image pixels and learn hash functions simultaneously.
- Sparse Matrix Based Hashing for Approximate Nearest Neighbor Search4
This paper proposes a new unsupervised hashing method based on a sparse matrix, named as Sparse Matrix based Hashing SMH, and demonstrates that SMH expedites the process of generating binary codes, and achieves competitive performance with the state-of-the-art un supervised hashing methods.
- 4
- Visual Topic Network: Building better image representations for images in social media4
This paper addresses the problem of building better image representations in social media, where the images are weakly annotated with user tags, and proposes a network structured topic model, namely Visual Topic Network (VTN), where the image content and their relations are simultaneously modeled.
- Beyond bag of words4
A novel image representation method in the sub-semantic space based representation of images that is not only more semantically meaningful but also more reliable and resistant to noise.
- 4
- 4
- Video Repeat Recognition and Mining by Visual Features4
This chapter presents methods to identify known and unknown video repeats respectively and addresses robust detection, searching efficiency and learning issues, and proposes a novel method to explore statistics of video database to estimate nearest neighbor classification error rate and learn the optimal classification threshold.
- 4
- Performance benchmarking of liver CT image segmentation and volume estimation4
A platform with necessary tools for performance benchmarking for algorithms of liver segmentation and volume estimation used for liver transplantation planning is presented, which includes an abdominal computer tomography (CT) image database (DB), annotation tools, a ground truth DB, and performance measure protocols.
- 4
- 4
- Mean shift based nonparametric motion characterization4
This paper employs the mean shift procedure and the histogram to propose a novel nonparametric motion representation and transforms the motion analysis to the classification problem of camera motion patterns in the presence of dominant object motion and non-dominant object motion.
- 4
- 4
- 3
- 3
- 3
- 3
- Category Aggregation Among Region Proposals for Object Detection3
A novel category aggregation approach based upon the observation that more frequently detected categories around an object have the higher probabilities to be present in an image, which can determine the most possible categories for an image in advance.
- Similarity-Preserving Binary Signature for Linear Subspaces3
The results show that the approximation with the binary signatures achieves an order of magnitude speed-up, while requiring significantly smaller amount of storage space, yet it still accurately preserves the similarity, and achieves high recognition accuracy comparable to the exact method in face recognition.
- Classification of Digital Rocks by Machine Learning3
An approach that the authors developed for classifying digital samples can be implemented to classify samples in progressive, recursive and regressive manners, and can be extended to develop correlation between the classes of samples and their fluid flow properties.
- Delineation of Liver Tumors from CT Scans Using Spectral Clustering with Out-of-Sample Extension and Multi-windowing3
An algorithm based on spectral clustering with out-of-sample extension with multi-windowing feature is developed for the semi-automated delineation of liver tumors from 3D CT scans with better results than using only the original data-depth and the support vector machine method.
- Object-Based Image Retrieval Beyond Visual Appearances3
Testing on Caltech-256 datasets shows that by fusing the visual words in a relevance consistent way, the visual synset can partially bridge visual differences of images of same class and deliver satisfactory retrieval of relevant images with different visual appearances.
- Combining Visual Features for Medical Image Retrieval and Annotation3
This paper's work using visual feature fusion for the tasks of medical image retrieval and annotation in the benchmark of ImageCLEF 2005 is reported, as one of the best performed runs.
- Stripe: Image Feature Based on a New Grid Method and Its Application in ImageCLEF3
A new image feature based on a new region division method of images, named ‘stripe’, which is proved by the applications in ImageCLEF’s medical subtasks, to be much faster at the calculation speed compared with other features.
- 3
- 3
- 3
- Uncover the balanced geometry in long-tailed contrastive language-image pretraining2
An improved method, termed as Geometry-Balanced CLIP (GeoCLIP), is proposed, which automatically constructs pseudo clusters and aligns them with a predefined equiangular geometric structure, thereby enjoying the theoretical merits of better maintaining the uniformity at the semantic level.
- 2
- CogniEdit: Dense Gradient Flow Optimization for Fine-Grained Image Editing2
A unified framework CogniEdit is proposed, combining multi-modal reasoning with dense reward optimization that propagates gradients across consecutive denoising steps, enabling trajectory-level gradient flow through the sampling process.
- 2
- 2
- 2
- 2
- Network Compression via Recursive Bayesian Pruning.2
This study explores, under Bayesian framework, a structured pruning method with layer-wise sequential dependency assumed, a more general learning setting, and derives a new dropout noise, which makes it possible to approximate the posterior of drop out noise knowing that of the previous layer.
- Creating Descriptive Visual Word Tree for Tag Ranking of Social Image2
Experimental results indicate that the proposed descriptive visual word tree can effectively improve the accuracy of tag ranking of social image as well as reduce the time of tagranking.
- 2
- Hybrid-Indexing Multi-type Features for Large-Scale Image Search2
A novel hybrid-indexing strategy, which incorporates multiple similarity metrics into one inverted index file during off-line indexing, which could obtain more satisfying retrieval results because the index file conveys hybrid similarities among images.
- Personalized Visual Vocabulary Adaption for Social Image Retrieval2
The personalized visual vocabulary adaption is proposed which removes non-discriminative visual words and replaces them with more exact and discriminative ones, i.e., adapt a general vocabulary toward a specific user's image set, and shows satisfying efficiency.
- 2
- Multi-order visual phrase for scalable partial-duplicate visual search2
Multi-order visual phrase (MVP) is proposed which contains two complementary clues: center visual word quantized from the local descriptor of each image keypoint and the visual and spatial clues of multiple nearby keypoints and is more robust to quantization error than existing visual phrases.
- Locality preserving verification for image search2
A new matching method based on spatial constraints that can not only estimate the number of correct matches between two images efficiently, but also determine the correctness of each match accurately, and is simple and easy to be implemented.
- 2
- Weakly supervised codebook learning by iterative label propagation with graph quantization2
A weakly supervised codebook learning framework, which integrates image labels to supervise codebook building with two steps: the Label Propagation step propagates image labels into local patches by multiple instance learning and instance selection and the Graph Quantization step integrates patch labels to build codebook using Mean Shift.
- Efficient l p -norm multiple feature metric learning for image categorization2
An efficient distance metric learning model which adapts Distance Metric Learning on multiple feature representations and is extended to nonlinear kernel learning and category specific metric learning, which demonstrate the applicability of using many existing kernels for image data and exploring the hierarchical semantic structures for large scale image datasets.
- Liver tumor detection and classification using content-based image retrieval2
The proposed method first detects liver abnormalities by eliminating the normal tissue/organ from the liver region, and in the second step it ranks these abnormalities with respect to spherical symmetry, compactness and size using a tumoroid measure to facilitate fast location of liver focal mass lesions.
- The 1st workshop on large-scale multimedia retrieval and mining (LS-MMRM'09)2
This workshop will provide a venue for the participants to explore a variety of aspects and applications on how advanced multimedia analysis techniques can be leveraged to address the challenges in large-scale data collections.
- 2
- FADA: An Efficient Dimension Reduction Scheme for Image Classification2
FADA can find a good projection with adaptation to different sample distributions and discover the classification in the subspace with naive Bayes classifier and also alleviates the overfitting problem implicitly caused by ADA.
- Ranking Metrics and Evaluation Measures2
A novel boosted distance metric is proposed that not only finds the best distance metric that fits the distribution of the underlying elements but also selects the most important feature elements in respect to similarity.
- Complete Performance Graphs in Probabilistic Information Retrieval2
This paper proposes a comprehensive and well normalized description of the ranking performance compared to the performance of an Ideal Retrieval System defined by ground-truth for a large number of predefined queries and advocates normalization with respect to relevant class size and restriction to specific normalized scope values.
- Nonparametric motion model2
This work presents a novel nonparametric motion model that bases the motion characterization on the classification of motion patterns and elegantly incorporates the spatial-range information for noise removal and discontinuity preserving smoothing of MVF.
- 1
- 1
- 1
- 1
- 1
- 1
- FTL: A Universal Framework for Training Low-Bit DNNs via Feature Transfer1
This paper introduces a novel feature-based knowledge transfer framework, which utilizes a 32-bit DNN to guide the training of a low- bit DNN via feature maps, and makes two branches interact via centered cosine distance rather than the widely-used p-norms.
- Editorial: Booming of Neural Networks and Learning Systems1
As you open this January issue of the IEEE Transactions on Neural Networks and Learning Systems (TNNLS), the author is very delighted and honored to report several key metrics of IEEE TNNLS to the community.
- 1
- Towards multi-modality invariance: A study in visual representation1
This chapter presents a comprehensive study in extracting invariant compact visual features, i.e., visual features that resist the changes in position, viewpoints, geometry, etc., serving as a basic of understanding video signals and assisting other modalities such as audio signals.
- Contextual modeling on auxiliary points for robust image reranking1
This paper proposes auxiliary points which are added as assistant neighbors in an unsupervised manner which act on revealing implicit similarity in the metric space and clustering matched image pairs and presents a contextual modeling on auxiliary points (CMAP) method for image reranking.
- 1
- 1
- Fast large-scale object retrieval with binary quantization1
This paper proposes a feature quantization algorithm called binary quantization, which converts a scale-invariant feature transform (SIFT) feature into a descriptive and discriminative bit-vector, which allows itself to adapt to the classic inverted file structure for box indexing.
- Indexing heterogeneous features with superimages1
This work proposes to package semantically relevant images into superimages, and then index superimages instead of single images, which naturally saves the memory cost and retrieval time and is naturally superior to many online retrieval fusion algorithms in terms of retrieval efficiency and memory consumption.
- 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
- Intelligent multimedia interactivity1
A method that given a query motion the logically relevant samples can be retrieved and one of the first attempts in using recent relevance feedback techniques for improving the action retrieval performance.
- Segmentation of liver and liver tumor for the Liver-Workbench1
A two-module image analysis procedure which integrates two novel semi-automatic algorithms has been developed to segment 3D liver and liver tumors from multi-detector computed tomography (MDCT) images.
- Visual topic model for web image annotation1
A novel Visual Topic Model framework is proposed, via unsupervised clustering techniques, that aims at analyzing image semantics fusing both content and context, by considering tag correlations and ambiguities.
- Semi-automatic photo clustering with distance metric learning1
This paper proposes a semi-automatic photo clustering scheme that attempts to improve clustering performance with users' interactions, and it is better than pure manual adjustment approach by exploring distance metric learning.
- Statistical learning for multimedia information retrieval and their applications1
This dissertation proposes a generalized Kernel Discriminant-EM algorithm (KDEM), which allows for a better classification in a nonlinear feature subspace, and combines Relevance Feedback and KDEM together to construct an efficient and effective semi-supervised learning framework.
- Generalized discriminant analysis in content-based image retrieval1
Novel discriminant analysis methods are generalized from the state-of-the-art techniques, such as Linear Discriminant Analysis (LDA), Principal Component Analysis (PCA) and Biased Discriminatory Analysis (BDA), to develop robust and efficient CBIR systems.
- Two-stage SVM for Medical Image Annotation.1
A two-stage medical image annotation method in order to achieve higher classification rate and significant improvement of classification accuracy has been achieved.
- A hybrid feature dimension reduction approach for image classification1
A novel hybrid feature dimension reduction techniqe is proposed to construct descriptive and discriminant features at the same time by maximizing the Rayleigh coefficient, which not only increases the effective dimension of the projected subspace, but also offers more flexibility and alternatives to LDA and PCA.
- –
- Application of Image Understanding and Human-Computer Interaction Technology in Intelligent Assisted Travel Systems–
The results indicate that image understanding-based human-computer interaction technology demonstrates high recognition accuracy and user satisfaction within intelligent assisted travel, significantly enhancing travel experience and service quality, thus providing valuable theoretical and practical references for the field of intelligent assisted travel.
- –
- –
- –
- –
- –
- –
- –
- See Blue Sky: Deep Image Dehaze Using Paired and Unpaired Training Images–
A cycle generative adversarial network is proposed to construct a novel end-to-end image dehaze model that could remove the haze of images and also restore the sky of images to be clean and blue (like captured in a sunny weather).
- –
- –
- –
- –
- –
- –
- Introduction to the Special Section on Intelligent Multimedia Systems and Technology Part II–
This section seeks effective machine-learning and data-mining algorithms, frameworks, systems, and implementations that particularly work on multimedia data (including image, video, and audio, which may also be associated with textual information).
- –
- Probabilistic indexing of media sequences–
A novel scheme to facilitate scalable and efficient media sequence retrieval based on advanced hashing algorithm using hidden Markov model (HMM) for comprehensive media sequence modeling and calculate HMM supervector to represent segments of media sequence.
- –
- –
- Content-Based Music Summarization and Classification–
This chapter covers the aspects of music summarization in compressed domain and uncompressed domain, music video summarization, music genre classification, and semantic region detection in acoustical music signals.
- Introduction to the special issue on intelligent multimedia systems and technology–
This special issue seeks effective machine-learning and data-mining algorithms, frameworks, systems, and implementations that particularly work on multimedia data (including image, video, and audio, which may also be associated with textual information).
- The use of non-conventional methods for content analysis and understanding–
This panel will enable the participants to understand key concepts, state-of-the-art techniques, and open issues in content analysis and understanding that make use of non-conventional methods.
- –
- –
- –
- Visualization, Estimation and User Modeling for Interactive Browsing of Personal Photo Libraries–
This chapter has developed a user-centric algorithm for visualization and layout for content-based image retrieval (CBIR) in large photo libraries and can adapt to the user’s notions of content, context and preferences in style and interactive navigation.
- –
- –
- –
- Voting based object boundary reconstruction–
An alternative approach based on tensor voting is introduced, which is applied to various video surveillance clips under different conditions, and the experimental results demonstrate significant improvement on the final extracted objects accuracy.
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.