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
- Yizhi LiuSuggested from co-authorship
- Jintao LiSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works242 from public data
- Automated pulmonary nodule detection in CT images using deep convolutional neural networks459
A novel automated pulmonary nodule detection framework with 2D convolutional neural network (CNN) is proposed to assist the CT reading process and it is illustrated that the proposed method can obviously achieve accurate pulmonary nodules detection.
- Read Like Humans: Autonomous, Bidirectional and Iterative Language Modeling for Scene Text Recognition450
This paper argues that the limited capacity of language models comes from: 1) implicitly language modeling; 2) unidirectional feature representation; 3) language model with noise input, and proposes an autonomous, bidirectional and iterative ABINet for scene text recognition.
- Frequency-aware Discriminative Feature Learning Supervised by Single-Center Loss for Face Forgery Detection355
A novel single-center loss (SCL) is designed that only compresses intra-class variations of natural faces while boosting inter-class differences in the embedding space and can learn more discriminative features with less optimization difficulty.
- Graph Structured Network for Image-Text Matching311
This paper presents a novel Graph Structured Matching Network (GSMN) to learn fine-grained correspondence of structured phrase correspondence, and shows that GSMN outperforms state-of-the-art methods on benchmarks.
- ContourNet: Taking a Further Step Toward Accurate Arbitrary-Shaped Scene Text Detection257
The ContourNet is proposed, which effectively handles the problems taking a further step toward accurate arbitrary-shaped text detection and suppresses false positives by only outputting predictions with high response value in both orthogonal directions.
- Curriculum Learning for Natural Language Understanding257
By reviewing the trainset in a crossed way, this work is able to distinguish easy examples from difficult ones, and arrange a curriculum for language models, and obtains significant and universal performance improvements on a wide range of NLU tasks.
- 248
- From Two to One: A New Scene Text Recognizer with Visual Language Modeling Network196
A Visual Language Modeling Network (VisionLAN), which views the visual and linguistic information as a union by directly enduing the vision model with language capability, which significantly improves the speed by 39% and adaptively considers the linguistic information to enhance the visual features for accurate recognition.
- Filtration and Distillation: Enhancing Region Attention for Fine-Grained Visual Categorization196
A novel “Filtration and Distillation Learning” (FDL) model is proposed to enhance the region attention of discriminate parts for FGVC that utilizes the proposing-predicting matchability as the performance metric of Region Proposal Network (RPN), thus enable a direct optimization of RPN to filtrate most discriminative regions.
- 183
- Domain-Aware Visual Bias Eliminating for Generalized Zero-Shot Learning181
A novel Domain-aware Visual Bias Eliminating (DVBE) network is proposed that constructs two complementary visual representations, i.e., semantic-free and semantic-aligned, to treat seen and unseen domains separately and outperforms existing methods by averaged 5.7% improvement.
- 176
- DEADiff: An Efficient Stylization Diffusion Model with Disentangled Representations129
This paper introduces DEADiff, a mecha-nism to decouple the style and semantics of reference images and attains the best visual stylization results and optimal balance between the text controllability inherent in the text-to-image model and style similarity to the reference image, as demonstrated both quantitatively and qualitatively.
- Semi-supervised User Profiling with Heterogeneous Graph Attention Networks110
The authors' heterogeneous graph attention networks (HGAT) method learns the representation for each entity by accounting for the graph structure, and exploits the attention mechanism to discriminate the importance of each neighbor entity.
- PRRNet: Pixel-Region relation network for face forgery detection106
A novel network, called Pixel-Region Relation Network (PRRNet), to capture pixel-wise and region-wise relations respectively for face forgery detection, which achieves the state-of-the-art detection performance on three recent proposed face forgeries detection datasets.
- Pediatric Seizure Prediction in Scalp EEG Using a Multi-Scale Neural Network With Dilated Convolutions105
This study proposes an end-to-end framework by using a temporal-spatial multi-scale convolutional neural network with dilated convolutions for patient-specific seizure prediction, providing a promising solution for EEG-based seizure prediction.
- Multi-Objective Matrix Normalization for Fine-Grained Visual Recognition96
Experiments demonstrate that the proposed Multi-Objective Matrix Normalization (MOMN) method is superior to existing normalization-based methods in terms of both accuracy and efficiency.
- ABINet++: Autonomous, Bidirectional and Iterative Language Modeling for Scene Text Spotting88
This paper argues that the limited capacity of language models comes from 1) implicit language modeling; 2) unidirectional feature representation; and 3) language model with noise input, and proposes an autonomous, bidirectional and iterative ABINet++ for scene text spotting.
- Convolutional Attention Networks for Scene Text Recognition81
The experimental results validate the effectiveness of different components and show that the convolutional-based method achieves state-of-the-art or competitive performance over prior works, even without the use of RNN.
- Robust Deep Co-Saliency Detection with Group Semantic79
A novel end-to-end deep learning approach for robust co-saliency detection by simultaneously learning highlevel group-wise semantic representation as well as deep visual features of a given image group by exploiting the inter-image interaction at semantic-level and the complementarity between group semantics and visual features.
- DiffAM: Diffusion-Based Adversarial Makeup Transfer for Facial Privacy Protection69
A novel face protection approach, dubbed DiffAM, is proposed, which leverages the powerful generative ability of diffusion models to generate high-quality protected face images with adversarial makeup transferred from reference images.
- 68
- Attention and Language Ensemble for Scene Text Recognition with Convolutional Sequence Modeling63
An attention-based architecture which is completely based on CNNs, and instead of using a single loss from language aspect, multiple losses from attention and language are accumulated for training the networks in an end-to-end way.
- 62
- 61
- Dynamic Inconsistency-aware DeepFake Video Detection61
This paper proposes a novel Dynamic Inconsistency-aware Network to handle the inconsistent problem, which uses a Cross-Reference module (CRM) to capture both the global and local inter-frame inconsistencies.
- 61
- Geometry Aligned Variational Transformer for Image-conditioned Layout Generation60
A novel paradigm entitled image-conditioned layout generation is explored, which aims to add text overlays to an image in a semantically coherent manner and proposes an Image-Conditioned Variational Transformer (ICVT) that autoregressively generates various layouts in an image.
- CARIS: Context-Aware Referring Image Segmentation59
Context-Aware Referring Image Segmentation (CARIS), a novel architecture that enhances the contextual awareness of linguistic features via sequential vision-language attention and learnable prompts, which achieves new state-of-the-art performances on three public benchmarks.
- 57
- Clinical application of artificial intelligence-assisted diagnosis using anteroposterior pelvic radiographs in children with developmental dysplasia of the hip57
The deep learning system was highly consistent, more convenient, and more effective for diagnosing DDH compared with clinician-led diagnoses and will improve the current artificially complicated screening referral process.
- 57
- 57
- SVTRv2: CTC Beats Encoder-Decoder Models in Scene Text Recognition56
SVTRv2, a CTC model endowed with the ability to handle text irregularities and model linguistic context, is proposed, and a feature rearrangement module is introduced to ensure that visual features accommodate the requirement of CTC, thus alleviating the alignment puzzle.
- Hierarchical multi-view context modelling for 3D object classification and retrieval55
A hierarchical multi-view context modelling method for 3D object classification and retrieval that can imitate the human need to look back and forth to identify and compare the discriminative parts of individual 3D objects based on a joint convolutional neural network and bidirectional long short-term memory network.
- PosterMaker: Towards High-Quality Product Poster Generation with Accurate Text Rendering52
This work identifies the key to precise text rendering as constructing a character-discriminative visual feature as a control signal and proposes a robust character-wise representation as control and develops TextRenderNet, which achieves a high text rendering accuracy of over 90%.
- 51
- 47
- 45
- Dynamic tissue analysis using time- and wavelength-resolved fluorescence spectroscopy for atherosclerosis diagnosis45
Simultaneous time- and wavelength-resolved fluorescence spectroscopy was developed and tested for the dynamic characterization of atherosclerotic tissue ex vivo and arterial vessels in vivo and demonstrates the potential of STWRFS as a tool for real-time optical characterization of arterial tissue composition and for atherosclerosis research and diagnosis.
- Detecting Tampered Scene Text in the Wild43
This paper introduces a new task, named Tampered Scene Text Detection (TSTD), to localize text instances and recognize the texture authenticity in an end-to-end manner and discusses the potential of frequency information for distinguishing feature learning, and pro-pose a parallel-branch feature extractor to enhance the feature representation capability.
- 41
- Residual domain dictionary learning for compressed sensing video recovery41
A temporal residual-domain dictionary learning method for CS video recovery that outperforms the state-of-art method near 1 dB in terms of peak signal noise rate (PSNR) at some higher sampling rate.
- Semantic-guided Reinforced Region Embedding for Generalized Zero-Shot Learning39
A novel Semantic-guided Reinforced Region Embedding (SR2E) network that can localize important objects in the long-term interests to construct semantic-visual embedding space and preserves the semantic relationship into visual features via semantic- visual alignment.
- Symmetrical Linguistic Feature Distillation with CLIP for Scene Text Recognition38
Overall, CLIP-OCR is the first to design a smooth transition between image and text for the STR task, and a new Linguistic Consistency Loss (LCL) is proposed to enhance the linguistic capability by considering second-order statistics during the optimization.
- 38
- Deep Cascaded Attention Network for Multi-task Brain Tumor Segmentation38
This paper proposes a simple yet efficient 3D deep cascaded attention network (DCAN) for brain tumor segmentation that is able to capture the hierarchical correlations of overlapping regions and simultaneously tackle multi-tasks in a single model.
- 36
- 35
- 33
- 32
- Wavelet-enhanced Weakly Supervised Local Feature Learning for Face Forgery Detection32
A novel wavelet-enhanced weakly supervised local feature learning framework is proposed and outperforms previous state-of-the-art methods when only image-level labels are available and achieves comparable or even better performance than counterparts using patch- level labels.
- 32
- Design, construction, and validation of a rotary multifunctional intravascular diagnostic catheter combining multispectral fluorescence lifetime imaging and intravascular ultrasound32
The feasibility of the catheter for simultaneous bimodal interrogation of arterial lumen and for generation of robust fluorescence lifetime data under IVUS guidance are demonstrated and facilitate further development of a FLIm-IVUS technique for intravascular diagnosis of atherosclerotic cardiovascular diseases including vulnerable plaques.
- 31
- 31
- Self-Supervised Attention Mechanism for Pediatric Bone Age Assessment With Efficient Weak Annotation31
- Neighborhood-Adaptive Structure Augmented Metric Learning30
This work proposes a Neighborhood-Adaptive Structure Augmented metric learning framework (NASA), where the neighborhood structure is realized as a structure embedding and learned along with the sample embedding in a self-supervised manner, and significantly improves and outperforms the state-of-the-art methods.
- CRNet: A Center-aware Representation for Detecting Text of Arbitrary Shapes30
This work proposes an anchor-free scene text detector leveraging Center-aware Representation to achieve accurate arbitrary-shaped scene text detection namely CRNet and proposes a center-aware location algorithm to explicitly learn center regions and center points of text instances, which is able to separate adjacent text instances effectively.
- DSRN: A Deep Scale Relationship Network for Scene Text Detection30
An end-to-end architecture to map multiscale convolution features onto a scale invariant space to obtain uniform activation of multi-size text instances and a novel Recall Loss is proposed to force the network to concern more about miss-detected text instances by up-weighting poor-classified examples.
- Choose What You Need: Disentangled Representation Learning for Scene Text Recognition, Removal and Editing29
This work synthesizes a dataset of image pairs with identical style but different content and proposes a Disentangled Representation Learning framework (DARLING), the first time in the field of scene text that disentangles the inherent properties of the text images.
- Balanced Classification: A Unified Framework for Long-Tailed Object Detection29
A novel foreground classification balance loss (FCBL) is developed to ameliorate the domination of head categories and shift attention to difficult-to-differentiate categories by introducing pairwise class-aware margins and auto-adjusted weight terms, respectively to prevent the over-suppression of tail categories in the context of unequal competition.
- Query-Memory Re-Aggregation for Weakly-supervised Video Object Segmentation27
A novel Re-Aggregation based framework, which uses feature matching to efficiently find the target and capture the temporal dependencies from multiple frames to guide the segmentation ofWeakly-supervised video object segmentation (WVOS).
- Dual Part Discovery Network for Zero-Shot Learning26
A novel Dual Part Discovery Network (DPDN) that considers both attribute and category discriminative information by discovering attribute- guided parts and category-guided parts simultaneously to improve knowledge transfer is proposed.
- ACE-Net: Biomedical Image Segmentation with Augmented Contracting and Expansive Paths26
An ACE-net is developed that aims to enhance the feature representation and utilization by augmenting the contracting and expansive paths by the recently proposed advanced techniques including ASPP, dense connection and deep supervision mechanisms, and novel connections such as directly connecting the raw image to the expansive side.
- Robust common visual pattern discovery using graph matching26
This paper formulate CVPs discovery as a graph matching problem, depending on pairwise geometric compatibility between feature correspondences, which consists of three components: Preliminary Initialization Optimization (PIO), Guided Expansion (GE) and Post Agglomerative Combination (PAC).
- 25
- 25
- Mask2DiT: Dual Mask-based Diffusion Transformer for Multi-Scene Long Video Generation24
Mask2DiT is proposed, a novel approach that establishes fine-grained, one-to-one alignment between video segments and their corresponding text annotations, and incorporates a segment-level conditional mask, which conditions each newly generated segment on the preceding video segments, thereby enabling auto-regressive scene extension.
- Hybrid-Level Instruction Injection for Video Token Compression in Multi-modal Large Language Models24
The Hybridlevel Instruction Injection Strategy for Conditional Token Compression in MLLMs (HICom), utilizing the instruction as a condition to guide the compression from both local and global levels to retain the maximum amount of user-focused information.
- 24
- Dynamic-Aware Federated Learning for Face Forgery Video Detection24
A Federated Learning face forgery detection framework to train a global model collaboratively while keeping data on local devices and achieves competitive performance compared with existing methods that are trained with centralized data, with higher-level security and privacy guarantee.
- Extract Bone Parts Without Human Prior: End-to-end Convolutional Neural Network for Pediatric Bone Age Assessment24
This paper proposes a novel single-stage Attention-Recognition Convolutional Neural Network (AR-CNN), believed to be the first end-to-end structure to extract bone parts for BAA without segmentation, detection and human prior.
- CA 3 Net24
A novel Contextual-Attentional Attribute-Appearance Network ($\rm CA^3Net) for person re-identification simultaneously exploits the complementarity between semantic attributes and visual appearance, the semantic context among attributes, visual attention on attributes as well as spatial dependencies among body parts, leading to discriminative and robust pedestrian representation.
- CAPability: A Comprehensive Visual Caption Benchmark for Evaluating Both Correctness and Thoroughness23
CAPability is introduced, a comprehensive multi-view benchmark for evaluating visual captioning across 12 dimensions spanning six critical views, and a heuristic metric, \textit{know but cannot tell} ($K\bar{T}$), indicating a significant performance gap between QA and caption capabilities.
- TextPainter: Multimodal Text Image Generation with Visual-harmony and Text-comprehension for Poster Design23
This study introduces TextPainter, a novel multimodal approach that leverages contextual visual information and corresponding text semantics to generate text images that can generate visually-and-semantically-harmonious text images for posters.
- Domain-Specific Embedding Network for Zero-Shot Recognition23
A novel Domain-Specific Embedding Network (DSEN) that can apply specific projections to different domains for unbiased embedding, as well as several domain constraints.
- Question-Aware Tube-Switch Network for Video Question Answering23
A novel Question- Aware Tube-Switch Network (TSN) for video question answering which contains a Mix module to synchronously combine the appearance and motion representation at time slice level and a Switch module to adaptively choose appearance or motion tube as primary at each reasoning step, guiding the multi-hop reasoning process.
- Dual-Stream Knowledge-Preserving Hashing for Unsupervised Video Retrieval22
This work decomposes video information into reconstruction-dependent and semantic-dependent information, which disentangles the semantic extraction from reconstruction constraint, and designs a simple dual-stream structure, including a temporal layer and a hash layer that naturally preserves the disentangled semantics into binary codes.
- 22
- Detecting Uyghur text in complex background images with convolutional neural network22
This paper develops an effective and efficient region-based convolutional neural network for Uyghur text detection in complex background images, which is much faster than the state-of-the-art methods while keeps competitive accuracy.
- 21
- Real-World Automatic Makeup via Identity Preservation Makeup Net21
This paper introduces a new makeup model, called Identity Preservation Makeup Net (IPM-Net), which preserves not only the background but the critical patterns of the original identity, and disentangle the face images to two different information codes, i.e., identity content code and makeup style code.
- Misshapen Pelvis Landmark Detection by Spatial Local Correlation Mining for Diagnosing Developmental Dysplasia of the Hip21
This paper investigates the spatial local correlation with convolutional neural network (CNN) for misshapen landmark detection and proposes a deep learning based method named FR-DDH network, which achieves an excellent precision in landmark location and illness diagnosis over human experts.
- TALK-Act: Enhance Textural-Awareness for 2D Speaking Avatar Reenactment with Diffusion Model20
The Motion-Enhanced Textural-Aware ModeLing for SpeaKing Avatar Reenactment (TALK-Act) framework, which enables high-fidelity avatar reenactment from only short footage of monocular video, and carefully construct 2D and 3D structural information as intermediate guidance.
- 20
- 20
- Extracting salient region for pornographic image detection20
A novel approach of extracting salient region for pornographic image detection by integrating a novel saliency map model with a skin-color model and a face detection model to capture ROI in pornographic images and a ROI-based codebook algorithm is proposed to enhance the representative power of visual-words.
- Learning Pixel Affinity Pyramid for Arbitrary-Shaped Text Detection19
A novel text detection framework is proposed, which jointly learns semantic segmentation and a pixel affinity pyramid in a unified fully convolutional network, which surpasses most existing text detectors and achieves state-of-the-art performance, denoting its superior capability in detecting arbitrary-shaped texts.
- 19
- 19
- Multi-Features Fusion and Decomposition for Age-Invariant Face Recognition19
This work proposes a novel Multi-Features Fusion and Decomposition (MFFD) framework to learn more discriminative feature representations and alleviate the intra-class variations for AIFR.
- 18
- 18
- Common visual pattern discovery via graph matching18
This paper formulate CVPs discovery as a graph matching problem, depending on pairwise geometric compatibility between feature correspondences, and proposes two algorithms--Preliminary Initialization Optimization (PIO) and Post Agglomerative Combining (PAC).
- 18
- 18
- 17
- Cluster and Scatter: A Multi-grained Active Semi-supervised Learning Framework for Scalable Person Re-identification17
A Multi-grained Active Semi-Supervised learning framework, named MASS, is proposed to address the scalable person re-identification problem existing in the practical scenarios and introduces a customized purification loss for the purified clustering, which utilizes the complementary information in both labeled and unlabeled data to optimize the model for solving the ineffectiveness problem.
- Fusing audio vocabulary with visual features for pornographic video detection17
This paper proposes a novel framework of fusing audio vocabulary with visual features for pornographic video detection, and shows that this approach outperforms the traditional one which is only based on visual features, and achieves satisfactory performance.
- 16
- 16
- Name-face association with web facial image supervision16
Methods for automatically associating faces detected from multimedia documents with their names presented in the surrounding metadata and an assigning-thresholding pipeline to uniformly ensure that the name-face correspondence is strictly one-to-one are described.
- Pairwise weak geometric consistency for large scale image search16
A novel pairwise weak geometric consistency constraint (P-WGC) method is presented, which utilizes the local similarity characteristic of deformations, and measures the pairwise geometric similarity of matches between two sets of local features.
- 15
- Multi-task hourglass network for online automatic diagnosis of developmental dysplasia of the hip15
An online diagnosis tool based on a multi-task hourglass network, which can accurately extract landmarks to detect the extent of hip dislocation and predict the age of the femoral head and improve the accuracy of the age prediction of femoral heads is established.
- 15
- Global context and boundary structure-guided network for cross-modal organ segmentation15
This paper innovatively employs the global context from all spatial regions to guide the deformable convolution, and design a novel loss that weights more on the errors of positions nearby organ boundaries, and this loss function can avoid errors brought by border blurs.
- STIDNet: Identity-Aware Face Forgery Detection With Spatiotemporal Knowledge Distillation14
It is argued that it is necessary to take both spatial and temporal perspectives into consideration for adequate inconsistency clues and proposed a novel forgery detector named SpatioTemporal IDentity network (STIDNet), which is more suitable for real-world implementation in terms of model complexity and reference set size.
- 14
- Boat in the Sky: Background Decoupling and Object-aware Pooling for Weakly Supervised Semantic Segmentation14
A background decoupling mechanism (BDM) is proposed to handle the semantic ambiguity by regularizing the consistency of predicted CAMs from the samples with identical foregrounds but different backgrounds and a global object-aware pooling is introduced to alleviate spatial ambiguity.
- Robust and parallel Uyghur text localization in complex background images14
A multi-color-channel enhanced maximally stable extremal region is used to extract components in images, which is robust to blur and low contrast and a component connected graph algorithm is proposed to construct text lines.
- 13
- 13
- 13
- Constructing Spatio-Temporal Graphs for Face Forgery Detection13
A novel forgery detector named Spatio-Temporal Graph Network (STGN) is proposed, which contains two kinds of graph-convolution-based units, the Spatial Relation Graph Unit (SRGU) and the Temporal Attention Graph unit (TAGU), which can combine the discriminative power of spatial inconsistency and the generalization capacity of temporal incoherence for face forgery detection.
- Masked Text Modeling: A Self-Supervised Pre-training Method for Scene Text Detection13
A novel self-supervised Masked Text Modeling (MTM) pre-training method for scene text detection, which can be trained with unlabeled data and improve the linguistic reasoning ability for text occlusion.
- 13
- End-to-end Boundary Exploration for Weakly-supervised Semantic Segmentation13
This paper argues that using a single model to accomplish image- and pixel-level classification will fall into the balance of multi-target and consequently weakens the recognition capability, and proposes an effective encoder-decoder framework to explore object boundaries and solve the above dilemma.
- Learning to Draw Text in Natural Images with Conditional Adversarial Networks13
This work designs a pixel-manipulation word generator constrained by self-regularization, which learns to convert local characters to plausible word image and proposes a novel style loss based on variance-minimization to achieve style consistency of characters.
- RegionRAG: Region-level Retrieval-Augmented Generation for Visual Document Understanding12
This work proposes RegionRAG, a novel framework that shifts the retrieval paradigm from the document level to the region level and enables the generator to focus solely on concise, query-relevant visual content, improving both efficiency and accuracy.
- Effective and Efficient Image Copy Detection Based on GPU12
A novel system based on Graphics Processing Units (GPU) that combines two complementary local features, Harris-Laplace and SURF, to provide a compact representation of an image to improve the accuracy and efficiency of image copy detection.
- Imaging apolipoprotein AI in vivo12
Using a modified apolipoprotein AI to create an HDL contrast agent provides a new approach to investigate HDL biodistribution, metabolism and regulation in vivo.
- 11
- Attention‐guided transformation‐invariant attack for black‐box adversarial examples11
To improve the transferability of adversarial examples, an attention‐guided transformation‐invariant adversarial attack method is proposed, which incorporates an attention mechanism to disrupt the most distinctive features and simultaneously ensures adversary attack invariance under different transformations.
- Look Back Again11
This paper proposes a dual parallel attention network (DPAN), in which a newly designed parallel context attention module (PCAM) is cascaded with the original PPAM, using linguistic contextual information to compensate for the information inconsistency between queries and keys.
- Adaptive Alignment Network for Person Re-identification11
A new Adaptive Alignment Network (AAN) is proposed, towards robust and accurate person re-identification, which automatically aligns pedestrian images from coarse to fine by learning both patch-wise and pixel-wise alignments, leading to effective pedestrian representation invariant to the variance of human pose and location across images.
- AlignZeg: Mitigating Objective Misalignment for Zero-Shot Semantic Segmentation10
AlignZeg markedly enhances zero-shot semantic segmentation, as shown by an average 3.8% increase in hIoU, primarily attributed to a 7.1% improvement in identifying unseen classes, and it is validated that the improvement comes from alleviating the objective misalignment issue.
- Deep Convolutional Nets for Pulmonary Nodule Detection and Classification10
A novel pulmonary nodule detection and classification system with 2D convolutional neural networks is proposed to effectively address the challenges in lung cancer diagnosis and early treatment and proves that the method can maintain a satisfactory sensitivity even with extremely low false positive rates.
- Igd: Instructional Graphic Design With Multimodal Layer Generatio9
Instructional Graphic Designer (IGD) is proposed to swiftly generate multimodal layers with editable flexibility with only natural language instructions and architecturally supports scalability and extensibility in complex graphic design tasks.
- 9
- 9
- 9
- 9
- 8
- REMOT: A Region-to-Whole Framework for Realistic Human Motion Transfer8
The REMOT adopts a progressive generation paradigm: it first generates each body part in the driving pose without flow-based warping, then composites all parts into a complete person of the driving motion to preserve the natural global appearance.
- CircleNet for Hip Landmark Detection8
A much simpler and more efficient framework called CircleNet is proposed to improve the accuracy of landmark detection by predicting landmark and corresponding radius and not only constrain the relationship between landmarks but also integrate landmark detection and object detection into an end-to-end framework.
- Application of deep-learning–based artificial intelligence in acetabular index measurement7
The newly constructed artificial intelligence system could quickly and accurately measure the acetabular index of standard anteroposterior pelvic radiographs and showed good credibility according to the Bland–Altman principle.
- Learning Rich Attention for Pediatric Bone Age Assessment7
It is shown it is possible to learn rich attention without the need for complicated network design or precise annotation – a simple module is all it takes.
- Temporal-Contextual Attention Network for Video-Based Person Re-identification7
A new Temporal-Contextual Attention Network (TCA-Net) for person re-identification in videos that exploits temporally local context among consecutive frames to concentrate selectively on crucial frames within a video sequence.
- Fast approximate matching of binary codes with distinctive bits7
A new clustering algorithm is proposed, which chooses cluster centers on the basis of relative distances and uses a more homogeneous partition of the dataset than HCT has to build the hierarchical clustering trees, and a new index is proposed using compressed binary codes based on hierarchical decomposition of binary spaces.
- Fast Search of Binary Codes with Distinctive Bits7
An algorithm to compress binary codes by extracting distinctive bits according to the standard deviation of each bit is proposed and a new index is proposed using com-pressed binary codes based on hierarchical decomposition of binary spaces.
- 6
- 6
- 6
- Hierarchical Granularity Transfer Learning6
A novel Bi-granularity Semantic Preserving Network (BigSPN) is proposed to bridge the granularity gap for robust knowledge transfer and experiments show that BigSPN is an effective framework for Hierarchical Granularity Transfer Learning.
- Triple-Bit Quantization with Asymmetric Distance for Image Content Security6
This paper proposes a novel triple-bit quantization strategy to solve the problem of retrieving relevant images from the large database by assigning 3-bit to each dimension and shows that the proposed method can achieve remarkable improvement in query accuracy compared to original binary embedding methods.
- 6
- Uyghur Language Text Detection in Complex Background Images Using Enhanced MSERs6
Experimental comparisons prove that the proposed algorithm is effective for detecting Uyghur language text in complex background images, much better than the state-of-the-art performance of 75.5%.
- RICS‐DFA: a space and time‐efficient signature matching algorithm with Reduced Input Character Set6
A character escaping and replacing scheme is first introduced to decrease the size of D FA's character set and then to reduce DFA's space requirement with a series of optimization techniques, and a RICS‐DFA constructing algorithm with time complexity of O(n) is presented.
- LRANet++: Low-Rank Approximation Network for Accurate and Efficient Text Spotting5
A novel parameterized text shape representation based on low-rank approximation for precise detection and a triple assignment detection head for fast inference is proposed and integrated into an end-to-end text spotting framework, termed LRANet++, capable of accurately and efficiently spotting arbitrary-shaped text.
- IDseq: Decoupled and Sequentially Detecting and Grounding Multi-Modal Media Manipulation5
An Image-Driven Decoupled Sequential Framework (IDseq), designed to decouple image features and rationally integrate them to accomplish different sub-tasks effectively, and a Decoupled Image Manipulation Decoder (DIMD) that processes image tasks within a decoupled schema.
- 5
- Meta semi-supervised medical image segmentation with label hierarchy5
This work proposes a meta-based semi-supervised segmentation framework with the exploitation of label hierarchy, and dynamically divide consistency regularization from supervised guidance as different domains, thereby bridging the knowledge gap.
- 5
- 5
- 5
- Adaptive Bilinear Pooling for Fine-grained Representation Learning5
This work proposes a generalized feature interaction method, named Adaptive Bilinear Pooling (ABP), which can adaptively infer a suitable pooling strategy for a given sample based on image content, and demonstrates the effectiveness of the proposed method on three widely used benchmarks.
- 5
- 5
- Video to Article Hyperlinking by Multiple Tag Property Exploration5
Both objective performance evaluations and subjective user studies show the effectiveness of the proposed hyperlinking, which produces more accurate contextual tags and thus a larger number of relevant articles than other approaches.
- Design, construction, and validation of a multimodal intravascular diagnostic catheter combining IVUS and fluorescence lifetime spectroscopy detection channels5
The feasibility of the described catheter for parallel interrogation of vessel walls based on TRFS and IVUS and to generate robust TRFS data are demonstrated and facilitate further development of a bi-modal TRFS-IVUS technique for intravascular diagnosis of atherosclerotic cardiovascular diseases including vulnerable plaques.
- Focus on the Whole Character: Discriminative Character Modeling for Scene Text Recognition4
A novel method that enriches the character features to enhance the discriminability of characters and achieves state-of-the-art performance on common benchmarks and Union14M-Benchmark.
- 4
- Exploiting Pre-Trained Language Models for Black-Box Attack against Knowledge Graph Embeddings4
This article proposes a novel black-box attack method that only requires access to knowledge graph data, making it more realistic in real-world attack scenarios and proposes to fine-tune the PLM model by enriching triple embeddings with structure information.
- 4
- Proxy Probing Decoder for Weakly Supervised Object Localization: A Baseline Investigation4
A novel method named Proxy Probing Decoder (PPD) is proposed, which utilizes the segmentation property of self-attention map in the self-supervised vision transformer and breaks through model fine-tuning with a novel proxy probing decoder.
- TDI TextSpotter: Taking Data Imbalance into Account in Scene Text Spotting4
This paper proposes a novel scene text spotter TDI that outperforms state-of-the-art methods on three public datasets in both text detection and end-to-end text recognition, which fully proves the effectiveness and robustness of the method.
- March on Data Imperfections: Domain Division and Domain Generalization for Semantic Segmentation4
A novel model-agnostic training optimization algorithm which has two prominent components: Domain Division and Domain Generalization to deal with data imbalance and to alleviate the degradation caused by label noise is presented.
- Improving Brain Tumor Segmentation with Dilated Pseudo-3D Convolution and Multi-direction Fusion4
This work decomposes 3D convolution to decouple volumetric information, in the same way human experts treat volume data, and proposes a Multi-direction Fusion module, which achieves competitive performance with other state-of-the-art methods on BraTS 2018 brain tumor segmentation task and is much more light-weight.
- Uyghur Text Matching in Graphic Images for Biomedical Semantic Analysis4
This paper proposes a system that extracts text from Uyghur biomedical images, and matches the text in a specific lexicon for semantic analysis, and proposes a GPU acceleration scheme for matching network to match a complete UYghur text line directly rather than a single window.
- Robust Uyghur Text Localization in Complex Background Images4
A multi-color-channel enhanced Maximally Stable Extremal Regions (MSERs) extraction scheme is used to capture text components robustly, and the strong classification and retrieve strategy (SCRS) accurately identifies text components.
- Data-Dependent Locality Sensitive Hashing4
A new data-dependent LSH algorithm, which has two-level structures to perform ANN search in high dimensional spaces, and results show that the search speed can be increased by 48 times compared to E2LSH, while keeping high search precision.
- 3
- 3
- ShowTable: Unlocking Creative Table Visualization with Collaborative Reflection and Refinement3
This work proposes ShowTable, a pipeline that synergizes MLLMs with diffusion models via a progressive self-correcting process, highlighting its effective multi-modal reasoning, generation, and error correction capabilities.
- Test-Time Scaling with Reflective Generative Model3
This work introduces the first reflective generative model MetaStone-S1, which obtains OpenAI o3-mini's performance via the new Reflective Generative Form, and provides a self-supervised process reward model which can directly learn the high-quality reasoning trajectory selection from the outcome reward.
- Proactive Deepfake Detection via Self-Verifiable Semantic Watermarking3
SVS-WM, a Self-Verifiable Semantic Watermarking strategy, to embed pairs of correlated watermarks within facial semantics, leveraging the inherent fragility of these features to enable robust Deepfake detection.
- 3
- 3
- Frequency-based Zero-Shot Learning with Phase Augmentation3
A novel ZSL method named Frequency-based Phase Augmentation (FPA) network is proposed, which learns an effective representation of the attributes in the frequency domain, and a Hybrid PhaseAugmentation (HPA) module is introduced to transform visual features into the frequencydomain and augment the phase component for better retention of semantic information of the Attributes.
- Dual Dynamic Proxy Hashing Network for Long-tailed Image Retrieval3
This work proposes a novel Dual Dynamic Proxy Hashing Network with two sets of learnable dynamic proxies, i.e. hash proxies and feature proxies, to improve the discrimination of hash codes for tail-class samples and introduces high-dimensional continuous feature proxies that can describe semantic relationships more precisely, contributing to hash codes learning.
- 3
- High Fidelity Face Swapping via Semantics Disentanglement and Structure Enhancement3
A novel Semantics and Structure-aware face swapping framework (S2Swap) that exploits semantics disentanglement and structure enhancement for high fidelity face generation and adaptively integrate semantics and structure information in a self-learning manner.
- RAIRNet: Region-Aware Identity Rectification for Face Forgery Detection3
A novel Region-Aware Identity Rectification Network (RAIRNet) is proposed to effectively rectify the identity bias and adaptively exploit the inconsistency local region and outperforms previous binary-based and identity-based methods.
- 3
- 3
- 3
- WaveCSN3
The results prove that the WaveCSN can improve 2.66 and 4.11 pixels at least on these two datasets compared to other methods, and achieves the state-of-the-art for landmark detection in hip X-ray images.
- Uyghur Text Localization with Fast Component Detection3
A fast framework for Uyghur text localization which handle above obstacles with two effective measures, a stroke-specific detector based candidate extraction scheme and a component similarity based clustering is raised, which neither need the component-level classification nor the extra computations.
- 3
- Context-Oriented Name-Face Association in Web Videos3
The novelty lies in the exploration of both visual content of the video and crowdsourced text of the context that aims to find more specific facial images from the Web to facilitate the association.
- 2
- IterMeme: Expert-Guided Multimodal LLM for Interactive Meme Creation with Layout-Aware Generation2
This work proposes IterMeme, an end-to-end interactive meme creation framework that utilizes a unified Multimodal Large Language Model (MLLM) to facilitate seamless collaboration among multiple components and introduces a parameter-shared dual-LLM architecture that decouples the intricate representations of reference images and text.
- 2
- From Evaluation to Defense: Advancing Safety in Video Large Language Models2
VideoSafety-R1 is proposed, a dual-stage framework achieving unprecedented safety gains through three innovations: the VideoSafetyThinking dataset contains 46k video-query-thinking response triplets, and Alarm Token-Guided Safety Fine-Tuning injects learnable alarm tokens into visual and textual sequences, enabling explicit harm perception across modalities via multitask objectives.
- 2
- Weakly Supervised Pediatric Bone Age Assessment Using Ultrasonic Images via Automatic Anatomical RoI Detection2
This work proposes a weakly supervised interpretable framework entitled USB-Net, utilizing ultrasonic pelvis images and only image-level age annotations, and provides the first ultrasonic BAA dataset composed of 1644 ultrasonic hip joint images with image- level labels of age and gender.
- 2
- Fast and scalable lock methods for video coding on many-core architecture2
The experimental results show that the locks with dedicated hardware support have higher performance than the software lock, and the distributed hardware lock is more scalable than the centralized hardware lock.
- 2
- 1
- THGS: Lifelike Talking Human Avatar Synthesis From Monocular Video Via 3D Gaussian Splatting1
A novel pipeline is proposed, THGS, that animates lifelike Talking Human avatars using 3D Gaussian Splatting (3DGS) and employs a body pose, expression and skinning weights joint optimization strategy to optimize these parameters on the fly, which aligns hand movements and expressions better with video input.
- 1
- 1
- Control-Talker: A Rapid-Customization Talking Head Generation Method for Multi-Condition Control and High-Texture Enhancement1
This paper introduces a two-stage method called Control-Talker to achieve rapid customization of identity in talking head model and high-quality generation based on multimodal conditions, and proposes a unified driving method for both audio and video, enabling the model to precisely control expressions, poses, and lighting under multi conditions.
- Learning Complete Topology-Aware Correlations Between Relations for Inductive Link Prediction1
This work proposes a novel subgraph-based method, namely TACO, to model Topology-Aware relations between relations that are highly correlated to their topological structures within subgraphs, and proves that semantic correlations between any two relations can be categorized into seven topological patterns.
- Triple-Bit Quantization with Asymmetric Distance for Nearest Neighbor Search1
A novel quantization strategy triple-bit quantization (TBQ) is proposed to solve the problem of nearest neighbor search by assigning 3-bit to each dimension and asymmetric distance algorithm is applied to re-rank candidates obtained from hamming space for the final nearest neighbors.
- RICS-DFA: Reduced Input Character Set DFA for Memory-Efficient Regular Expression Matching1
A compact, time-efficient and novel DFA structure to significantly decrease the DFA’s space, the new DFA called Reduced Input Character Set DFA (RICS-DFA) is proposed, which reduces the memory consumption by 68 %–92 % while sacrificing trivial matching speed.
- A Novel Approach for Region-of-Interest Detection in Pornographic Images1
A novel saliency map model is firstly constructed, then it is integrated with a skin-color model and a face detection model to capture ROI in pornographic images more precisely than using the skin- color model alone.
- Motion Controlled Text-to-4D Generation with Multi-View Video Diffusion Model–
Motion4D is proposed, a rigging-free framework that animates static humanoid meshes with arbitrary initial poses by replicating the motion pattern of a reference sequence by adopting a generate-then-animate paradigm and introducing a Gaussian-Proxied Mesh Animation algorithm.
- –
- –
- –
- –
- –
- Few-shot bone age assessment via contrast relation adaptation–
A parallel learning strategy is used where the network learns from MiG data while simultaneously acquiring knowledge from the MaG data, and an in-batch ranking loss is designed that leverages the contrast relationships of MaG as an optimization objective in parallel learning, thus facilitating the MiG task as a supplementary constraint.
- Anomaly identification method of medium- and long-term electricity consumption information under multihop reasoning of knowledge graph–
A method for identifying abnormal medium and long-term electricity consumption information under multi-hop reasoning of knowledge graphs is designed, using the Z-score method to make a preliminary judgment of point anomalies in historical electricity consumption data.
- –
- –
- –
- –
- –
- Attention-driven frequency-based Zero-Shot Learning with phase augmentation–
A novel frequency-based framework that transforms spatial features into the frequency domain, allowing for more robust attribute representation and improved noise suppression and establishes a new direction in frequency-based zero-shot learning.
- –
- –
- –
- –
- –
- Law Is Order: Protecting Multimedia Network Transmission by Game Theory and Mechanism Design–
A novel Media-attribute Switching (MAS) is proposed, where the network allocate the transmission resources according to the application’s claimed media-attribute, thus to protect the multimedia network transmission.
- –
- Supervised deep hashing for image content security–
A supervised deep hashing method is proposed, which seeks multiple non-linear transformations to generate more discriminative binary codes with short bits and achieves very competitive results with the state-of-the-arts.
- Uyghur language text detection in images–
Experimental comparisons on the proposed algorithm prove that the algorithm is effective for detecting Uyghur Language text in complex background images, much better than the state-of-the- art performance.
- Hierarchical Encoding of Binary Descriptors for Image Matching–
This paper hierarchically encode the intensity differences between pixel pairs to build a discriminative binary feature and shows that the encoded descriptor is as efficient as the state-of-the-art binary descriptors in both computing and matching.
- Performance Analysis of Digital Tracking Loops for Telemetry Ranging Applications–
A practical algorithm to compute the loop parameters for discrete update (DU) and continuous update (CU) loop formulations is described and it is shown that a simple power-series approximation to the DU model is valid over a large range of time-bandwidth product.
- –
- Research of model retrieval algorithm based on VM eigenvalue combination–
It is proved that the model retrieval algorithm based on the VM eigenvalue combination could greatly improve the retrieval accuracy and efficiency in a small professional model library though the experiment.
- –
- Model Retrieval Based on Three-View–
A new 3D shape retrieval method based on three views that leads to an accurate matching of sequences even in the presence of local shifting on the shape.
- A Routing Protocol with Network Life Time and Other Network Performance Balancing Based on Ant Colony Optimization for Mobile Ad Hoc Networks–
A routing protocol for mobile Ad Hoc networks called Ant-Based Energy-Aware Routing Protocol (ABEAR), which is based on ant colony optimization (ACO), outperforms AODV greatly in terms of life time, packet delivery ratio and average end-to-end delay.
- –
- Targeting Apolipoproteins in Magnetic Resonance Imaging–
The physical property variation arises from each lipoprotein’s distinct composition, which varies from the low-density chylomicron to the high-density HDL, which has less cholesterol, triacylglycerol, and phosphorus but more protein.
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 2026-10-15. 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.