Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks140 from public data
- 496
- 454
- Selective-Supervised Contrastive Learning with Noisy Labels258
Experiments on multiple noisy datasets demonstrate the robustness of the learned representations by the proposed selective-supervised contrastive learning (Sel-CL), following the state-of-the-art performance.
- Low-Resolution Face Recognition in the Wild via Selective Knowledge Distillation210
This paper proposes a learning approach to recognize low-resolution faces via selective knowledge distillation by selectively distil the most informative facial features from the teacher stream by solving a sparse graph optimization problem, which is then used to regularize the fine-tuning process of the student stream.
- Bootstrapping Multi-View Representations for Fake News Detection149
A novel scheme of Bootstrapping Multi-view Representations (BMR) for fake news detection and improves Multi-gate Mixture-of-Expert networks (iMMoE) are proposed for feature refinement and fusion.
- 133
- The Ninth Visual Object Tracking VOT2021 Challenge Results127
The Visual Object Tracking challenge VOT2021 is the ninth annual tracker benchmarking activity organized by the VOT initiative; results of 71 trackers are presented; many are state-of-the-art trackers published at major computer vision conferences or in journals in recent years.
- Student Network Learning via Evolutionary Knowledge Distillation114
Instead of a fixed pre-trained teacher, an evolutionary teacher is learned online and consistently transfers intermediate knowledge to supervise student network learning on-the-fly to improve the transfer effectiveness of teacher knowledge.
- Detecting Deepfake Videos with Temporal Dropout 3DCNN111
A Temporal Dropout 3-Dimensional Convolutional Neural Network (TD-3DCNN) is proposed to detect deepfake videos and can enhance the representation and generalization ability, avoiding model overfitting and improving detecting accuracy.
- Estimating Noise Transition Matrix with Label Correlations for Noisy Multi-Label Learning95
This work studies the identifiability problem of the class-dependent transition matrix in noisy multi-label learning, and inspired by the identifiability results, a new estimator is proposed by exploiting label correlations without neither anchor points nor accurate class posterior, leading to better classification performance.
- 82
- M3D: Dataset Condensation by Minimizing Maximum Mean Discrepancy80
A novel DM-based method named M3D for dataset condensation is presented by Minimizing the Maximum Mean Discrepancy between feature representations of the synthetic and real images by embedding their distributions in a reproducing kernel Hilbert space, resulting in a more generalized condensed set.
- Accurate Temporal Action Proposal Generation with Relation-Aware Pyramid Network79
In RapNet, a novel relation-aware module is introduced to exploit bi-directional long-range relations between local features for context distilling and generates superior accurate proposals over the existing state-of-the-art methods.
- Efficient Low-Resolution Face Recognition via Bridge Distillation78
By learning low- resolution face representations and mimicking the adapted high-resolution knowledge, a light-weight student model can be constructed with high efficiency and promising accuracy in recognizing low-resolution faces.
- Look Through Masks: Towards Masked Face Recognition with De-Occlusion Distillation78
This work proposes to migrate the mechanism of amodal completion for the task of masked face recognition with an end-to-end de-occlusion distillation framework, which consists of two modules that applies a generative adversarial network to perform face completion.
- 76
- Predicting Aesthetic Score Distribution Through Cumulative Jensen-Shannon Divergence74
A novel CNN based on the Cumulative distribution with Jensen-Shannon divergence (CJS-CNN) is presented to predict the aesthetic score distribution of human ratings, with a new reliability-sensitive learning methodbased on the kurtosis of the score distribution, which eliminates the requirement of the original full data of human rating data (without normalization).
- WebUAV-3M: A Benchmark for Unveiling the Power of Million-Scale Deep UAV Tracking67
A fine-grained UAV tracking-under-scenario constraint (UTUSC) evaluation protocol and seven challenging scenario subtest sets are constructed to enable the community to develop, adapt and evaluate various types of advanced trackers.
- Deepfake Video Detection via Predictive Representation Learning64
A predictive representative learning approach termed Latent Pattern Sensing to capture semantic change characteristics for deepfake video detection, which can simultaneously describe the latent patterns of videos across frames spatially and temporally in a unified way, leading to an effective and robust deep fake video detector.
- ILGNet: inception modules with connected local and global features for efficient image aesthetic quality classification using domain adaptation62
A novel deep convolutional neural network named ILGNet is proposed, which combines both the inception modules and a connected layer of both local and global features in order to label an input image as high- or low-aesthetic quality.
- Deepfake Video Detection with Spatiotemporal Dropout Transformer60
This paper proposes a simple yet effective patch-level approach to facilitate deepfake video detection via spatiotemporal dropout transformer that reorganizes each input video into bag of patches that is then fed into a vision transformer to achieve robust representation.
- Aesthetic Attributes Assessment of Images55
The evaluation criteria used in image captions is used to prove that the specially designed AMAN model outperforms traditional CNN-LSTM model and modern SCA-CNN model of imageCaptions.
- 53
- 50
- 44
- 41
- 3D textured model encryption via 3D Lu chaotic mapping40
This work proposes a 3D Lu chaotic mapping based encryption method that can encrypt and decrypt 3D textured models correctly and typical statistic and brute-force attacks can be resisted by the proposed method.
- Trustable Co-Label Learning From Multiple Noisy Annotators39
This paper proposes a data-efficient approach, called Trustable Co-label Learning (TCL), to learn deep classifiers from multiple noisy annotators when a small set of trusted data is available, which effectively uses trusted data as a guide to generate trustable soft labels (termed co-labels).
- Distilling Channels for Efficient Deep Tracking39
This paper presents a novel and general framework termed channel distillation to facilitate deep trackers and demonstrates that an integrated formulation can turn feature compression, response map generation, and model update into a unified energy minimization problem to adaptively select informative feature channels that improve the efficacy of tracking moving objects on the fly.
- Federated Learning with Label-Masking Distillation38
This paper focuses on label distribution skew in federated learning, where due to the different user behavior of the client, label distributions between different clients are significantly different, and proposes a label-masking distillation approach termed FedLMD to facilitate Federated learning via perceiving the various label distributions of each client.
- 37
- Talking Face Generation with Expression-Tailored Generative Adversarial Network35
An end-to-end Expression-Tailored Generative Adversarial Network (ET-GAN) to generate an expression enriched talking face video of arbitrary identity and the results show the effectiveness of the ET-GAN in generating high quality expressional talking face videos against existing state-of-the-arts.
- Look One and More: Distilling Hybrid Order Relational Knowledge for Cross-Resolution Image Recognition33
A teacher-student learning approach to facilitate low-resolution image recognition via hybrid order relational knowledge distillation, which preserves the similarities in various order relational structures, leading to a better knowledge transfer.
- Fast cascade face detection with pyramid network33
This paper proposes a fast CNNs cascade face detector with multi-task learning and network acceleration techniques, which is comparable in performance with the-state-of-arts while the speed reaches astonishing 165 fps on Titan GPU.
- Coupled-View Deep Classifier Learning from Multiple Noisy Annotators32
To alleviate the propagation of incorrect labels, small-loss metric is proposed to select reliable instances in both views and the approach can finally learn a robust data classifier which less overfits to label noise.
- View-specific subspace learning and re-ranking for semi-supervised person re-identification31
A View-Specific Semi-supervised Subspace Learning (VS-SSL) approach that can learn specific projections for each view, utilizing limited labeled data to guide the training while leveraging abundant unlabeled data simultaneously is put forward.
- Low-Resolution Object Recognition With Cross-Resolution Relational Contrastive Distillation30
This study proposes a cross-resolution relational contrastive distillation approach to facilitate low-resolution object recognition that enables the student model to mimic the behavior of a well-trained teacher model which delivers high accuracy in identifying high-resolution objects.
- 30
- Image editing by object-aware optimal boundary searching and mixed-domain composition30
This paper proposes a general approach, called object-aware image editing, to obtain consistency in structure, color, and texture in a unified way, and improves upon previous gradient-domain composition in three ways.
- Color image encryption in non-RGB color spaces30
The experimental results reveal that the encryption methods in Non-RGB color spaces can achieve similar results as the method that conducts the same encryption level in each channel of the RBG color space, including the resistance to several attacks such as brute-force attack, statistic attack, correlation attack, while consuming less time.
- Transferring Annotator- and Instance-Dependent Transition Matrix for Learning From Crowds28
Theoretical analyses are derived to demonstrate that both the knowledge transfer from global to individuals and the knowledge transfer between neighboring individuals can effectively help mitigate the challenge of modeling general AIDTM.
- 28
- 27
- Learning Privacy-Preserving Student Networks via Discriminative-Generative Distillation27
This work proposes a discriminative-generative distillation approach to learn privacy-preserving deep models, taking models as bridge to distill knowledge from private data and then transfer it to learn a student network via two streams.
- An Image Encryption Algorithm Based on Zigzag Transformation and 3-Dimension Chaotic Logistic Map26
An image encryption algorithm based on Zigzag transformation and 3-Dimension Logistic chaotic map by making use of the permutation-diffusion encryption structure is proposed, which has high key sensitivity, high encryption speed and the strong ability to resist exhaustive attack and statistical attack.
- Adaptive Wavelet Filter With Edge Compensation for Remote Sensing Image Denoising25
An adaptive wavelet filter for image denoising is proposed that could effectively reduce the variable stripe noise from different observation targets and an edge compensation method, fitted to the scanning model of the AGRI, is suggested to avoid ringing artifacts.
- Fewer-Shots and Lower-Resolutions25
This paper proposes a few-shot knowledge distillation approach to learn an ultrafast face recognizer via two steps and shows that the proposed approach performs favorably against state-of-the-arts in recognizing low-resolution faces with an extremely low memory of 30KB.
- 23
- 22
- Differentially Private Learning with Grouped Gradient Clipping21
A differentially private deep learning paradigm to train private models is presented and a simple operation termed grouped gradient clipping is proposed to modulate the gradient weights to provide with strong privacy protection and avoid accuracy degradation, providing a good trade-off between privacy and performance.
- 20
- Proposal pyramid networks for fast face detection19
Results on several public benchmarks show that the proposed face detector achieves comparable accuracy to the-state-of-arts while the detection speed reaches 60 FPS with an i5 CPU, which significantly exceeds previous DCNN-based face detectors.
- Accurate UAV Tracking with Distance-Injected Overlap Maximization18
This work is improving the accuracy of target localization by deriving a conceptually simple target localization loss and a global feature recalibration scheme in a mutual reinforced way, and its minimal solution can alleviate the drift problem caused by camera motion.
- Traffic Sign Recognition Based on Attribute-Refinement Cascaded Convolutional Neural Networks18
Experiments show that the proposed hierarchical cascaded framework can extract the deep information of similar categories, improve discrimination of the model and increase classification accuracy of traffic signs.
- Latent Pattern Sensing: Deepfake Video Detection via Predictive Representation Learning15
This work proposes a predictive representative learning approach termed Latent Pattern Sensing to capture semantic change characteristics for deepfake video detection, which cascades a CNN-based encoder, a ConvGRU-based aggregator and a single-layer binary classifier.
- Secure face retrieval for group mobile users14
This paper proposes a protocol of privacy preserving face retrieval in the cloud for mobile users, which protects the user photos and the face detector simultaneously and can successfully retrieve the proper photos from the cloud server and protect the user photos and the face detector.
- Ultrafast Video Attention Prediction with Coupled Knowledge Distillation12
An extremely light-weight network with ultrafast speed, named UVA-Net, constructed based on depth-wise convolutions and takes low-resolution images as input, which is comparable to 11 state-of-the-art models in video attention prediction and costs only 0.68 MB memory footprint.
- 12
- 11
- 11
- ScratchNet: Detecting the Scratches on Cellphone Screen11
An automated scratches detection method by cascading two main modules that applies a lightweight CNN model, ScratchNet, to identify each small scratch candidate whether it is real scratch or not, which outperforms the LeNet model and the other classifiers.
- Color Image Encryption in CIE L*a*b* Space11
This paper is the first time to encrypt color image in CIE L*a*b* color space and the method can resistant several attacks such as brute-force attack, statistic attack, correlation attack.
- Compressing and Accelerating Neural Network for Facial Point Localization10
A novel method is proposed to compress and accelerate large DNN models while maintaining the performance, which achieves 26 × compression and 4 × acceleration while the root mean squared error (RMSE) increases by just 3.6%.
- 9
- 9
- 8
- 7
- 7
- Structure injected weight normalization for training deep networks6
The proposed deep structural weight normalization methods can reduce the number of trainable parameters while guaranteeing high accuracy, whereas DSWN-NM can accelerate the convergence while improving the performance of deep networks.
- Occluded Face Recognition by Identity-Preserving Inpainting.6
Experimental results prove the proposed Identity Preserving Generative Adversary Networks capable of dealing with varieties of occlusions and producing photorealistic and identity-preserving results, promoting occluded face recognition performance.
- 6
- 6
- Predicting Aesthetic Radar Map Using a Hierarchical Multi-task Network6
A hierarchical multi-task dense network is proposed to make multiple regression of the properties of images to obtain sufficient aesthetic information of the image, which is of certain guiding significance to the comprehensive evaluation of image aesthetics.
- 6
- PKI: Prior knowledge-infused neural network for few-shot class-incremental learning5
This work retains more prior knowledge and proposes a prior knowledge-infused neural network (PKI) to facilitate FSCIL, and designs two variants of the prior knowledge-infused neural network to trade off a balance between resource consumption and performance by reducing the number of projectors.
- 5
- 5
- 5
- Defending Against Adversarial Examples via Soft Decision Trees Embedding5
This paper extracts the semantic features of adversarial examples with a feature extraction network and embeds a specific soft decision tree embedded to select the key semantic features for each feature map from convolutional layers and the selected features are fed to a light-weight classification network.
- Single reference image based scene relighting via material guided filtering5
A novel outdoor scene relighting method which needs only a single reference image and is based on material constrained layer decomposition is proposed, which can produce similar illumination effect as that of the reference image on the input image using only asingle reference image.
- Efficient Deep Aesthetic Image Classification using Connected Local and Global Features5
A novel DCNN structure codenamed ILGNet is proposed for image aesthetics classification, which introduces the Inception module and connects intermediate Local layers to the Global layer for the output and makes the aesthetic assessment model more easily to be integrated into mobile and embedded systems.
- 5
- 4
- 4
- 4
- Personalized Federated Learning via Backbone Self-Distillation4
A backbone self-distillation approach to facilitate personalized federated learning that involves learning two components: the shared backbone for common representation and the private head for local personalization, which enables effective global knowledge transfer.
- Privacy-Preserving Student Learning with Differentially Private Data-Free Distillation4
This paper presents an effective teacher-student learning approach to train privacy-preserving deep learning models via differentially private data-free distillation, and proposes a label differential privacy algorithm termed selective randomized response to protect the label information.
- Scene Relighting Using a Single Reference Image Through Material Constrained Layer Decomposition4
A novel outdoor scene relighting method, which needs only a single reference image and is based on material constrained layer decomposition, which can produce similar illumination effect as that of the reference image on the input image using only asingle reference image.
- 4
- 4
- 3
- 3
- Double-Blinded Finder: a two-side secure children face recognition system3
Double-Blinded Finder can provide efficient face matching while protecting the privacies of both sides: (1) the suspicious missing children side for avoiding the invasion of the human rights, and (2) the true missing children side for preserving the secondary victimization.
- 3
- Geometry-Based Global Alignment for GSMS Remote Sensing Images3
A geometry-based global alignment method to align landmarks and non-landmarks in the GSMS images with high accuracy, recall and significantly low computation complexity is proposed.
- 3
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 1
- Multi-Label Noise Transition Matrix Estimation with Label Correlations: Theory and Algorithm1
This paper investigates the identifiability of class-dependent transition matrices in noisy multi-label learning and proposes a novel estimator that leverages label correlations without the need for anchor points or precise fitting of noisy class posteriors, leading to excellent classification performance.
- Generative image inpainting with neural features1
An image inpainting approach based on generative adversarial networks (GANs) that can make good use of the features information of images, and perform efficient and realistic inPainting is proposed.
- Extracting Features of Interest from Small Deep Networks for Efficient Visual Tracking1
This paper proposes a lightweight deep tracker to facilitate efficient tracking by using a small deep convolutional neural network, which adaptively extracts features of interest from different deep layers for representing different objects, and then integrates into discriminative correlation filter formulation.
- 1
- 1
- 1
- 1
- –
- –
- –
- –
- Privacy-Preserving Model Transcription With Differentially Private Synthetic Distillation–
This work presents privacy-preserving model transcription, a data-free model-to-model conversion solution to facilitate model deployment with a privacy guarantee, and theoretically proves that this approach can guarantee differential privacy and convergence.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- Blind Face Retrieval for Mobile Users–
This paper proposes a protocol of privacy preserving face retrieval in the cloud for mobile users, which protects the user photos and the face detector simultaneously, and test the protocol inside several families and classes reveals that it can successfully retrieve the proper photos from the cloud server and protect theuser photos andThe face detector.
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
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.