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
- Suggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works151 from public data
- Deformable Convolutional Networks7,148
This work introduces two new modules to enhance the transformation modeling capability of CNNs, namely, deformable convolution and deformable RoI pooling, based on the idea of augmenting the spatial sampling locations in the modules with additional offsets and learning the offsets from the target tasks, without additional supervision.
- 1,270
- Face Liveness Detection from a Single Image with Sparse Low Rank Bilinear Discriminative Model586
This paper develops two new extensions to the sparse logistic regression model which allow quick and accurate spoof detection and proposes two strategies to extract the essential information about different surface properties of a live human face or a photograph, in terms of latent samples.
- RESOUND: Towards Action Recognition Without Representation Bias429
Experimental evaluation confirms the effectiveness of RESOUND to reduce the static biases of current datasets.
- Instance-Sensitive Fully Convolutional Networks413
This paper develops FCNs that are capable of proposing instance-level segment candidates that do not have any high-dimensional layer related to the mask resolution, but instead exploits image local coherence for estimating instances.
- 306
- 174
- Simultaneous estimation of image quality and distortion via multi-task convolutional neural networks172
- Beyond Physical Connections: Tree Models in Human Pose Estimation157
This paper surprisingly finds that no latent variables are introduced in the Leeds Sport Dataset (LSP) during learning latent trees for deformable model, which aims at approximating the joint distributions of body part locations using minimal tree structure.
- 136
- GRAPE: Generalizing Robot Policy via Preference Alignment128
GRAPE enhances the performance of state-of-the-art VLA models, increasing success rates on in-domain and unseen manipulation tasks by 51.79% and 58.20%, respectively, and can be aligned with various objectives, such as safety and efficiency.
- Free-hand sketch recognition by multi-kernel feature learning103
This work proposes a Multiple Kernel Learning (MKL) framework for sketch recognition, fusing several features common to sketches, and investigates the use of attributes as a high-level feature for sketches and shows how this complements low-level features for improving recognition performance under the MKL framework.
- 97
- 95
- 92
- A Survey of Deep Facial Attribute Analysis72
A general pipeline that deep facial attribute analysis follows, which comprises two stages: data preprocessing and model construction is summarized and the underlying theories of this two-stage pipeline for both FAE and FAM are introduced.
- A New Algorithm for Detecting Text Line in Handwritten Documents70
This work model text line detection as an image segmentation problem by enhancing text line structure using a Gaussian window, and adopting the level set method to evolve text line boundaries.
- Arbitrary Talking Face Generation via Attentional Audio-Visual Coherence Learning65
A novel arbitrary talking face generation framework is proposed by discovering the audio-visual coherence via the proposed Asymmetric Mutual Information Estimator (AMIE) and a Dynamic Attention (DA) block by selectively focusing the lip area of the input image during the training stage, to further enhance lip synchronization.
- 63
- 62
- VALHALLA: Visual Hallucination for Machine Translation59
A visual hallucination framework, called VALHALLA, which requires only source sentences at inference time and instead uses hallucinated visual representations for multi-modal machine translation, and demonstrates the effectiveness over both text-only baselines and state-of-the-art methods.
- Heterogeneous Attention Network for Effective and Efficient Cross-modal Retrieval57
This work proposes a heterogeneous attention network (HAN) for effective and efficient cross-modal retrieval, and proposes a new soft-max triplet loss, which adaptively gives more attention to harder negatives and thus trains the proposed HAN in a more effective manner compared with the original doublet loss.
- Anti-Makeup: Learning A Bi-Level Adversarial Network for Makeup-Invariant Face Verification52
This paper proposes a learning from generation approach for makeup-invariant face verification by introducing a bi-level adversarial network (BLAN), and achieves state-of-the-art verification accuracy across makeup status and can produce photo-realistic non-makeup face images.
- A convolutional neural networks denoising approach for salt and pepper noise51
Experimental results show that NLSF-CNN outperforms the state-of-the-art denoising algorithms with a few training images.
- 51
- 50
- Random Bits Forest: a Strong Classifier/Regressor for Big Data50
Testing with datasets from the UCI (University of California, Irvine) Machine Learning Repository shows that RBF outperforms other popular methods in both accuracy and robustness, especially with large datasets.
- Unsupervised Image Deraining: Optimization Model Driven Deep CNN45
This work discovers a simple yet important domain knowledge that directional rain streak is anisotropic while the natural clean image is isotropic, and designs an optimization model driven deep CNN in which the unsupervised loss function of the optimization model is enforced on the proposed network for better generalization.
- Coupled adversarial learning for semi-supervised heterogeneous face recognition39
A coupled adversarial learning (CAL) approach for the VIS-NIR face matching is proposed by performing adversarial learn on both image and feature levels to reduce the spectrum domain discrepancy and the over-fitting problem.
- Pose-preserving Cross Spectral Face Hallucination39
This work presents an approach to avert the data misalignment problem and faithfully preserve pose, expression and identity information during cross-spectral face hallucination and outperforms current state-of-the-art HFR methods at a high resolution.
- Combo-Attention Network for Baidu Video Advertising36
This paper introduces a technique used in Baidu video advertising for feeding relevant video ads according to the user's query, and proposes a Combo-Attention Network (CAN), built upon the proposed combo-attention module, which exploits cross-modal attentions besides self attentions to effectively capture the relevance between words and bounding boxes.
- Learning a bi-level adversarial network with global and local perception for makeup-invariant face verification35
Experimental results demonstrate that the proposed learning from generation approach for makeup-invariant face verification by introducing a bi-level adversarial network achieves state-of-the-art verification accuracy across makeup status and can produce photo-realistic non-makeup face images.
- 32
- Image Classification and Restoration of Ancient Textiles Based on Convolutional Neural Network30
A reliable retrieval and restoration system is designed to realize the repair of damaged textile images, reduce the difficulty of repair, and help users retrieve and browse different categories of ancient textile images, thus solve the problems of slow retrieval speed in traditional retrieval methods and poor restoration effect of ancient textile images.
- 28
- 26
- 26
- 25
- Disentangled Representation Learning of Makeup Portraits in the Wild22
Both visual and quantitative experimental results on four makeup datasets demonstrate the superiority of the proposed disentangled feature learning strategy, which proposes a cosmetic loss to learn makeup styles in a delicate way.
- 21
- 21
- Joint sparse coding based spatial pyramid matching for classification of color medical image19
Experimental results show that JScSPM provides significant improvements as compared with the majority voting based ScSPM and the original Sc SPM for color medical image classification.
- 19
- 18
- 17
- 17
- 16
- RGB-D salient object detection via cross-modal joint feature extraction and low-bound fusion loss14
A novel RGB-D salient object detection method via cross-modal joint feature extraction and low-bound fusion loss, designed to constrain the predictions of the two streams, to improve the lower bound of saliency values and generate a distinct saliency map.
- 14
- Multi-modal Dictionary BERT for Cross-modal Video Search in Baidu Advertising13
This paper presents the recent work for video advertisement retrieval, Multi-modal Dictionary BERT (MDBERT) model, which integrates a joint dictionary, which is shared among video features and word features and fosters effective cross- modal attention.
- EGM: Enhanced Graph-based Model for Large-scale Video Advertisement Search12
The enhanced graph-based model is enhanced through sub-path embedding to differentiate similar videos, beneficial for modeling fine-grain details to discriminate similar items.
- 12
- 12
- A deep learning based assisted analysis approach for Sjogren’s syndrome pathology images11
The study demonstrated deep learning’s potential to analysis pathology images, offering a reference framework for the application of deep learning technology in the medical domain.
- Eucalyptus Plantation Area Extraction Based on SLPSO-RFE Feature Selection and Multi-Temporal Sentinel-1/2 Data11
A feature selection method that combines multi-temporal Sentinel-1 and Sentinel-2 data with SLPSO (social learning particle swarm optimization) and RFE (Recursive Feature Elimination) and significantly elevates the classification precision of eucalyptus plantations across all three classifiers.
- Informative Sample Mining Network for Multi-domain Image-to-Image Translation10
It is revealed that improving the sample selection strategy is an effective solution for multi-domain image-to-image translation, and a novel multi-stage sample training scheme to reduce sample hardness while preserving sample informativeness is proposed.
- HAGAN: Hybrid Augmented Generative Adversarial Network for Medical Image Synthesis9
The hybrid augmented generative adversarial network (HAGAN) is proposed to maintain the authenticity of structural texture and tissue cells and achieves state-of-the-art performance in both high-resolution and low-resolution.
- 9
- Assorted Attention Network for Cross-Lingual Language-to-Vision Retrieval9
An assorted attention network (A2N) is proposed to synchronously overcome the language gap, bridge the modal gap and fuse features of two modals in an elegant and effective manner in the cross-lingual language-to-vision retrieval task.
- 9
- A Dynamic YOLO-Based Sequence-Matching Model for Efficient Coverless Image Steganography8
This work presents an efficient coverless scheme based on dynamically matched substrings that outperforms most previous works on data load, transmission security, and hiding capacity.
- Let's Play Music: Audio-Driven Performance Video Generation8
A multi-staged framework to generate realistic and synchronized performance video from given music to provide both global appearance and local spatial information by generating the coarse videos and keypoints of body and hands from a given music respectively.
- 8
- 8
- Detection of Self-explosive Insulators in Aerial Images Based on Improved YOLO v47
A self-explosive insulator detection algorithm based on the improved YOLO v4 is proposed that has a mAP of 92.0 % on the self- Exploding insulator dataset and has an F1-score of 78.70% on TT100K dataset, which has positive significance for the research on self-Explosive Insulator detection.
- Crack segmentation network based on hybrid-window transformer and dual-branch fusion6
A crack segmentation network based on hybrid-window Transformer and dual-branch fusion (CSHD) that can effectively capture local texture details and global context modeling to achieve high-precision crack segmentation.
- 6
- 6
- 6
- Learning Visual Shape Lexicon for Document Image Content Recognition6
This work presents a new approach for document image content categorization using a lexicon of shape features that is capable to handle high intra-class variability and shows results that exceed other state-of-the-art approaches, allowing it to be used as a content recognizer in image indexing and retrieval systems.
- Adapting Network Information Into Semantics for Generalizable and Plug-and-Play Multi-Scenario Network Diagnosis5
An LLM-based Network Semantic Generation (LNSG) algorithm, which integrates semanticization and symbolization methods to uniformly represent multimodal network information, is proposed, which enables rapid adaptation to diverse network environments and provides efficient fault diagnosis capabilities.
- 5
- 5
- 5
- 5
- 5
- End-to-End Light Field Image Compression with Multi-Domain Feature Learning4
A novel multi-domain feature learning-based light field image compression network (MFLFIC-Net) is proposed to improve compression efficiency and achieves superior performance on MS-SSIM and PSNR metrics compared to public state-of-the-art methods.
- A Two-Stage Real Image Deraining Method for GT-RAIN Challenge CVPR 2023 Workshop UG$^{\textbf{2}}$+ Track 34
The overall framework is elaborately designed and able to handle both heavy rainy and foggy sequences provided in the final testing phase and rank 1st on the average structural similarity (SSIM) and rank 2nd on theaverage peak signal-to-noise ratio (PSNR).
- 4
- 4
- 4
- Learning discriminative representations for semantical crossmodal retrieval4
Two effective methods are proposed to mine semantic correlations in crossmodal semantic features that can be retrieved across different modalities based on a flexible “feature up-lifting and down projecting” mechanism.
- Human articulated body recognition method in high-resolution monitoring images4
A novel formulation for detecting articulated human bodies from high-resolution monitoring images using a coarse detection process to detect the human object patches from monitoring images and an accelerated and improved salient mask is used to improve the performance of the dense scale-invariant feature transform descriptor.
- 4
- 4
- 3
- 3
- 3
- 3
- Multimodal image fusion network with prior-guided dynamic degradation removal for extreme environment perception2
The experimental results demonstrate that the proposed method significantly outperforms mainstream methods on multiple public datasets and in degraded scenarios such as smog, low light, and overexposure, demonstrating superior performance in terms of image clarity and quantitative metrics.
- 2
- Enhanced Multi-Modal Heart Segmentation Approach based on Swin-Transformer for Small Datasets2
An enhanced multi-modal ventricular segmentation method based on the U-Net network and vision Transformer model is proposed to address limitations of both the full convolutional network, which struggles with learning long-distance feature relationships, and the traditional vision Transformer model with its high computational load.
- Lightweight face anti-spoofing method based on cross-fused multi-modal features2
A lightweight face anti-spoofing method based on cross-fused multi-modal features was proposed, and the lightweight network ShuffleNetV2 is improved, thereby improving the classification accuracy while reducing the number of model parameters.
- 2
- 2
- Tree-based Text-Vision BERT for Video Search in Baidu Video Advertising2
A tree-based combo-attention network (TCAN) which has been recently launched in Baidu’s dynamic video advertising platform is presented, which provides a practical solution to deploy the heavy cross-modal attention for the large-scale query-to-video search.
- 2
- Ubiquitous Fitting: Ontology-Based Dynamic Exercise Program Generation2
This research aims to develop an ontology-driven knowledge-based system to dynamically generating personalized exercise programs exposing REST style web services, which can be accessed from any Internet-enabled device and deployed in cloud computing environments.
- 2
- Enhancing image quality for single-pixel imaging with a score-based diffusion model1
This work proposes a novel approach using the score-based diffusion model, leveraging the strong and stable generative capabilities of the diffusion model to fit the underlying distribution of target images and outperforms previous methods in terms of PSNR or SSIM at different sampling rates.
- Application and optimization of virtual simulation teaching assisted by artificial intelligence in ideological and political courses in universities1
The results of the study show that this new teaching mode can improve students' interest and participation in learning, deepen their understanding and application of ideological and political theories, and has important practical and theoretical value for improving the teaching quality and effect of the Civic and Political Science courses in colleges.
- Zeitgebers-Based User Time Perception Analysis and Data-Driven Modeling via Transformer in VR1
The RSTC method reveals that underestimating time in VR is strongly associated with improved user experience, presence, and engagement, and TPM-Net shows potential for modeling time perception in VR, enabling inference of relative changes in users' time perception and corresponding changes in user experience.
- 1
- Text-to-Classic: A Diffusion Method for Classical Art Generation Based on Text1
This paper proposes a lightweight diffusion model Text-to-Classic(T2C) to synthesize classical art paintings according to text descriptions and shows that the method can achieve good performance with fewer parameters.
- 1
- 1
- 1
- 1
- 1
- 1
- Heterogeneous Facial Analysis and Synthesis1
This chapter gives an overview of the heterogeneous problem in facial analysis as well as its synthesis solution, followed by a brief outline of the rest chapters.
- 1
- 1
- Consensus of rankings.1
The proposed multi-facet approach, without the need for pairwise comparison between rankings, allows to measure the consensus in a set of rankings, with respect to the length of common patterns, the number of commonpatterns for a given length, and thenumber of all common patterns.
- Visual enhancement based on salient region detection and layered difference representation1
Experimental results including comparisons with other methods show that the proposed algorithm produces more suitable enhanced images compared with the results of existing algorithms.
- 1
- 1
- A MAC-Layer Retransmission Algorithm Designed for Zigbee Protocol1
An ESCW-based Zigbee protocol performs significantly better - in terms of energy usage, throughput, and the time to complete the data collection task - than standard Zigbee (IEEE802.15.4).
- Natural Language-Powered Functional Protein Sequence and Structure Co-Design with Multi-Modal Knowledge Fusion–
This work proposes a Protein sequence–structure–function Consistency Design model empowered by Natural Language function description, dubbed ProtcdNl, which expands the protein design space and ensures alignment with function-aware framework, achieving the design of novel, diverse proteins with high functional fidelity.
- –
- –
- Multi-Source Unsupervised Graph Domain Adaptation via Concise Propagation–Transformation Pipeline–
This paper proposes a novel approach, which leverages Concise Propagation–Transformation pipeline for multi-source unsupervised Graph Domain Adaptation, dubbed as CPT-GDA, to better capture complementary knowledge from multiple sources in an efficient manner.
- –
- Efficient Hi-Fi Style Transfer via Statistical Attention and Modulation–
A novel feature modulation mechanism based on parameterized normalization, where the modulation parameters for content and style features are learned using a dual convolution network (BiConv), improves both the stability and quality of the style transfer process.
- –
- Doubly Debiased Test-Time Prompt Tuning for Vision-Language Models–
A dynamic retrieval-augmented modulation module that retrieves high-confidence knowledge from a dynamic knowledge base using the test image feature as a query, and uses the retrieved knowledge to modulate the predictions, and develops a reliability-aware prompt optimization module that incorporates a confidence-based weighted ensemble and cross-modal consistency distillation to impose regularization constraints during prompt tuning.
- –
- –
- –
- RoNet: Rotation-oriented Continuous Image Translation–
A novel rotation-oriented solution to model the continuous generation with an in-plane rotation over the style representation of an image, achieving a network named RoNet, and design a patch-based semantic style loss that learns the different styles of the similar object in different domains.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Enhanced multi-view prediction structure–
Experimental results show that the proposed method can achieve better performance than the previous schemes.
- –
- –
- –
- The Summarizing and Judging of Research on Generative Teaching–
The research of generative teaching in foreign countries is in progress through the meaning of pedagogy and psychology, but different problems exist in the research.
- Security Mechanism of ASP.NET and Its Implementation in Network Courseware–
This paper analyzes the security mechanism of ASP.NET such as system configuration, authentication, authorization and information encryption, which is applied in the development process of distance educational network courseware of 《Computer Graphics Foundation》, and a safe result of webpage applications has obtained.
- –
- –
- Application research on Kinescope-feedback Control in the Serving Technique Teaching of Elective Volleyball Courses–
Based on the mechanism of formation and the timely feedback of action skills, adopted kinescope-feedback controlling method, this research aims to provide a proper learning circumstance of "pursuing for knowledge" and "exploration" in order to inspire students' study motion and thinking.
- –
- –
- A morphological method for moving object segmentation and posture recognition–
A coarse-to-fine morphological segmentation and matching method, which uses a multiscale morphological image filter and a “spore” element, which is robust enough and can be easily parallelized.
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.