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
- Stan Ziqing LiSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works140 from public data
- GiT: Graph Interactive Transformer for Vehicle Re-Identification178
A graph interactive transformer (GiT) is proposed, which is a newly-designed local correction graph, which learns discriminative local features within a patch by exploring nodes’ relationships and is superior to state-of-the-art vehicle re-identification approaches.
- Vehicle Re-Identification Using Quadruple Directional Deep Learning Features158
The proposed QD-DLF approach outperforms multiple state-of-the-art vehicle re-identification methods and is applicable to both VeRi and VehicleID databases.
- 135
- 134
- Exploring Spatial Significance via Hybrid Pyramidal Graph Network for Vehicle Re-Identification126
Extensive experiments on three large scale vehicle databases demonstrate that the proposed HPGN is superior to state-of-the-art vehicle re-identification approaches in terms of accuracy, parameter cost, and computation cost.
- Multiple improved residual networks for medical image super-resolution116
The proposed multiple improved residual network (MIRN) super-resolution reconstruction method can better reconstruct the details and textures of images and avoid the over-smoothing of medical images after iteration, and the performance of the proposed algorithm is revealed to be better than that of existing state-of-the-art methods.
- Super-resolution reconstruction of knee magnetic resonance imaging based on deep learning113
The experimental results show that the EMISR framework has improved performance and greatly reduces the number of parameters and training time and provides a more powerful medical analysis in knee osteoarthritis examinations.
- Multi-label convolutional neural network based pedestrian attribute classification107
The proposed multi-label convolutional neural network (MLCNN) can simultaneously predict multiple pedestrian attributes and significantly outperforms the SVM based method on the PETA database.
- 102
- A Benign and Malignant Breast Tumor Classification Method via Efficiently Combining Texture and Morphological Features on Ultrasound Images86
The proposed method obtains a 91.11% accuracy, a 94.34% sensitivity, and an 86.49% specificity, which outperforms many related benign and malignant breast tumor classification methods.
- Deep Hybrid Similarity Learning for Person Re-Identification84
A deep hybrid similarity learning (DHSL) method based on a convolution neural network that is able to reasonably assign complexities of feature learning and metric learning in a CNN, so that the performance of person Re-ID is improved.
- 79
- 60
- A multi-order derivative feature-based quality assessment model for light field image60
The proposed MDFM is more consistent with the perception of the HVS on the evaluation of LFI than the classical and state-of-the-art IQA methods.
- 55
- 52
- HOG-assisted deep feature learning for pedestrian gender recognition51
Extensive experiments have shown that the proposed HDFL method is able to effectively recognize the pedestrian gender, and consistently outperforms the state-of-the-art methods.
- 42
- 41
- 40
- 39
- Joint Feature and Similarity Deep Learning for Vehicle Re-identification38
The proposed JFSDL method applies a siamese deep network to extract deep learning features for an input vehicle image pair simultaneously and is superior to multiple state-of-the-art vehicle re-identification methods on both the VehicleID and VeRi data sets.
- A multi-scale contrast-based image quality assessment model for multi-exposure image fusion37
Simulation results have clearly shown that the proposed multi-scale contrast-based model (MCM) model is more consistent with the perception of the HVS on the evaluation of MEF images than multiple state-of-the-art IQA methods.
- 35
- 35
- 34
- Multi-label learning with multi-label smoothing regularization for vehicle re-identification32
The proposed MLL with MLSR approach can effectively improve the performance delivered by the baseline and outperform multiple state-of-the-art vehicle re-ID methods as well.
- Progressively Hybrid Transformer for Multi-Modal Vehicle Re-Identification31
Experimental results show the proposed PHT method wins the state-of-the-art method by a larger 2.7% m AP on RGBNT100 and a larger 6.6% mAP on RGBN300, demonstrating that the proposed method can learn multi-modal complementary information effectively.
- Perceptual feature guided rate distortion optimization for high efficiency video coding31
By incorporating the proposed method into the HEVC, extensive simulation results have demonstrated that the proposed approach can significantly improve the perceptual coding performance and obtain better visual quality of the reconstructed video, compared with the original RDO in HEVC.
- 29
- 29
- 28
- 25
- 24
- An interpretive constrained linear model for ResNet and MgNet24
A constrained linear data- feature-mapping model as an interpretable mathematical model for image classification using a convolutional neural network (CNN) is proposed and a general data-feature iterative scheme is proposed to demonstrate the rationality of MgNet.
- Viewpoint robust knowledge distillation for accelerating vehicle re-identification24
Experiments show that the proposed VRKD method outperforms many state-of-the-art vehicle re-identification approaches with better accurate and speed performance.
- 24
- Towards a Smaller Student: Capacity Dynamic Distillation for Efficient Image Retrieval23
A Capacity Dynamic Distillation framework is proposed, which constructs a student model with editable representation capacity, initially a heavy model to fruitfully learn distilled knowledge in the early training epochs, and the stu-dent model is gradually compressed during the training.
- 22
- 22
- 22
- VRSDNet: vehicle re-identification with a shortly and densely connected convolutional neural network22
This paper solves Vehicle Re-identification with a Shortly and Densely connected convolutional neural Network (VRSDNet) with extensive experiments on both VeRi and VehicleID datasets show that the proposed VRSDNet is obviously superior to multiple state-of-the-art vehicle re-Identification methods in terms of accuracy and speed.
- 21
- Attribute-Image Person Re-identification via Modal-Consistent Metric Learning21
This paper proposes a modal-consistent metric learning (MCML) method that stably measures comprehensive similarities between attributes and images and integrates the CMMT loss function and MCMR, requiring no complex cross-modal feature alignments.
- 21
- 20
- 19
- Visual Attention Guided Pixel-Wise Just Noticeable Difference Model19
Experimental results indicate that compared with existing relevant JND profiles, the proposed JND model tolerates more distortion in the same perceptual quality, and brings better visual perception in the the same level of the injected JND-noise energy.
- 18
- 18
- 17
- Cross-modal group-relation optimization for visible–infrared person re-identification16
This paper explores a cross-modal group-relation (CMGR) to describe the relationship between the same group of people in two different modalities, and designs a group-relation correlation (GRC) loss function based on Pearson correlations to optimize CMGR.
- 16
- 15
- 14
- 14
- 14
- 14
- RASP: Regularization-based Amplitude Saliency Pruning13
A regularization-based amplitude saliency pruning criterion (RASP) is proposed that not only attains data independence but also establishes norm criteria for usage guidelines, and model saliency and an adaptive parameter group lasso approach sensitive to different layers are introduced.
- A Challenging Benchmark of Anime Style Recognition13
A cross-role protocol to evaluate ASR performance, in which query and gallery images must come from different roles to validate an ASR model is to learn abstract painting style rather than learn discriminative features of roles, is designed.
- Object Re-identification Using Teacher-Like and Light Students13
This work proposes a joint distillation and pruning (JDP) method to learn teacher-like and light (TLL) students for object Re-ID and designs a pruner-convolution-pruner block to replace a K × K convolutional layer of the student network.
- 13
- 12
- 11
- Improve Pedestrian Attribute Classification by Weighted Interactions from Other Attributes11
This work proposes a novel pedestrian attribute classification method which exploits interactions among different attributes and is able to keep the balance of the independent decision score and interaction of other attributes to yield more robust classification results.
- 10
- Deep-Learned and Hand-Crafted Features Fusion Network for Pedestrian Gender Recognition10
The proposed DHFFN method is superior to the state-of-the-art pedestrian gender recognition methods and the feature fusion process that is to combine these two features together for fully exploring the merits from both deep-learned and HOG features.
- 9
- 9
- Memory-augmented shuffled meta learning for visible–infrared person re-identification8
This work introduces a novel approach called memory-augmented shuffled meta (MASM) learning that not only enhances data utilization but also empowers the model to learn comprehensive global meta metrics, significantly improving its ability to distinguish individuals across modalities.
- 7
- 7
- 7
- 7
- 6
- Unsupervised video-based action recognition using two-stream generative adversarial network6
An unsupervised video-based action recognition method, called two-stream generative adversarial network (TS-GAN), which comprehensively learns the static texture and dynamic motion information inherited in videos with taking the detail information and global information into account is proposed.
- 6
- Joint Pyramid Feature Representation Network for Vehicle Re-identification6
Extensive experimental results on two commonly-used vehicle databases show that the proposed JPFRN is superior to multiple recently-developed vehicle Re-ID methods.
- 6
- 6
- 5
- 5
- 5
- 5
- 5
- 5
- A spatial structural similarity triplet loss for auxiliary vehicle re-identification5
where Xi, j,k and Yi,j,k are n×n sized patches of X and Y lied at i-th row, j-th column and k-th channel, which are located by pixel-by-pixel sliding over the k- fourth channel feature map.
- 5
- 5
- 4
- 4
- 4
- CNNs with Compact Activation Function4
It is shown that MgNet and ResNet with hat function still exhibit a slightly better generalization performance than CNNs with ReLU function by the experiments of classification on MNIST, CIFAR10/100 and ImageNet datasets, indicating that CNNs without spectral bias can have a good generalization capability.
- 3
- 3
- 3
- 3
- 3
- 2
- 2
- 2
- Focusing on pedestrians like human for clothes changing person re-identification2
This paper proposes a humanoid focus-inspired image augmentation (HFIA) method, which is intuitive image processing rather than a sophisticated network architecture designed to enhance local nuances of pedestrian images.
- 2
- 2
- 2
- Vehicle Re-identification Using Joint Pyramid Feature Representation Network2
The proposed JPFRN method obtains four vehicle feature blocks with different depths by designing pyramidal feature fusion of each convolution block in a basic network and is validated on a commonly used vehicle database.
- 2
- Multi-task learning network for handwritten numeral recognition2
The determination of scratchy/non-scratchy in the writing style learning module can effectively assist the handwritten numeral learning module to obtain a more robust and distinguishable feature so as to improve the recognition performance.
- 2
- 2
- UDD: Unsupervised denoising diffusion for noisy multi-focus image fusion1
This work proposes an unsupervised denoising diffusion (UDD) framework for noise-robust fusion without relying on noise-specific training, and introduces three key components: an unsupervised pixel-space noise-aware conditional diffusion (NACD) that introduces a pixel-space conditional formulation, eliminating the need for explicit degradation matrices and their transformed-domain decomposition.
- 1
- A federated vehicle re-identification benchmark and performance optimization using dual-phase contrastive dynamic aggregation1
A dual-phase contrastive dynamic aggregation (DCDA) method is proposed, which adjusts aggregation weights based on model changes during training, allocating greater weights during significant early changes and smaller weights later, allowing clients with smaller datasets to effectively influence the model.
- 1
- Global Instance Relation Distillation for convolutional neural network compression1
A Global Instance Relation Distillation (GIRD) for convolutional neural network compression is proposed, which improves both the instance-level and relation-level globality and outperforms state-of-the-art approaches in various visual tasks.
- 1
- 1
- 1
- 1
- Face Detection Based on Multi-block Quad Binary Pattern1
A novel local texture descriptor, called multi-block quad binary pattern (MB-QBP), is proposed, which has more features to conduct a better training process to refine the classifier and the over-fitting problem becomes much smaller in the MB-Q BP-based classifier.
- –
- –
- –
- Uncertainty-driven Progressive Single Image De-raining–
The proposed Uncertainty-driven Progressive Single Image De-raining (UPSID) outperforms multiple state-of-the-art algorithms, including single-stage, progressive, and uncertainty-driven single image de-raining methods.
- –
- NA-MMIF: a noise-agnostic multi-modal image fusion framework–
NA-MMIF consistently outperforms state-of-the-art fusion methods and denoise-then-fuse pipelines, particularly under challenging noise conditions, and is substantially outperforming the best competing fusion method by 19.5% and 9.9%, respectively.
- Align What Truly Matters: Pedestrian‐Relevant Hierarchical Parsing Network for Text‐Based Person Retrieval–
This paper proposes a pedestrian‐relevant hierarchical parsing (PHP) module to extract well‐aligned fine‐grained visual and textual features for alignment, and designs an expert‐driven feature parsing (EFP) module, which integrates the representational power of a mixture of experts with a modality‐aware gating mechanism to uncover deep semantic associations between text and image features.
- –
- –
- –
- –
- –
- Semantic segmentation in power grid scenarios using scale-transforming transformer–
The approach’s novelty lies in a scale-transforming module (STM), which implements cost-effective aspect ratio adjustments, patch splitting, and patch combining, and which seamlessly integrates into existing transformer models to build STT models.
- –
- –
- –
- –
- –
- –
- –
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 2026-10-16. 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.