Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks377 from public data
- Benchmarking Single-Image Dehazing and Beyond2,349
This work presents a new large-scale benchmark consisting of both synthetic and real-world hazy images, called REalistic Single-Image DEhazing (RESIDE), which highlights diverse data sources and image contents, and is divided into five subsets, each serving different training or evaluation purposes.
- Generalizing to Unseen Domains: A Survey on Domain Generalization1,906
This paper provides a formal definition of domain generalization and discusses several related fields, and categorizes recent algorithms into three classes: data manipulation, representation learning, and learning strategy, and presents several popular algorithms in detail for each category.
- An End-to-End Spatio-Temporal Attention Model for Human Action Recognition from Skeleton Data1,071
This work proposes an end-to-end spatial and temporal attention model for human action recognition from skeleton data on top of the Recurrent Neural Networks with Long Short-Term Memory (LSTM), which learns to selectively focus on discriminative joints of skeleton within each frame of the inputs and pays different levels of attention to the outputs of different frames.
- Image-adaptive watermarking using visual models993
This work proposes perceptually based watermarking schemes in two frameworks: the block-based discrete cosine transform and multiresolution wavelet framework and discusses the merits of each one, which are shown to provide very good results both in terms of image transparency and robustness.
- Co-Occurrence Feature Learning for Skeleton Based Action Recognition Using Regularized Deep LSTM Networks930
This work takes the skeleton as the input at each time slot and introduces a novel regularization scheme to learn the co-occurrence features of skeleton joints, and proposes a new dropout algorithm which simultaneously operates on the gates, cells, and output responses of the LSTM neurons.
- The Sixth Visual Object Tracking VOT2018 Challenge Results804
The Visual Object Tracking challenge VOT2018 is the sixth annual tracker benchmarking activity organized by the VOT initiative; results of over eighty trackers are presented; many are state-of-the-art trackers published at major computer vision conferences or in journals in the recent years.
- View Adaptive Recurrent Neural Networks for High Performance Human Action Recognition from Skeleton Data675
A novel view adaptation scheme to automatically regulate observation viewpoints during the occurrence of an action by design a view adaptive recurrent neural network with LSTM architecture, which enables the network itself to adapt to the most suitable observation viewpoints from end to end.
- A Twofold Siamese Network for Real-Time Object Tracking672
The proposed SA-Siam outperforms all other real-time trackers by a large margin on OTB-2013/50/100 benchmarks and proposes a channel attention mechanism for the semantic branch.
- Relation-Aware Global Attention for Person Re-Identification651
This work proposes an effective Relation-Aware Global Attention (RGA) module which captures the global structural information for better attention learning and proposes to stack the relations, i.e., its pairwise correlations/affinities with all the feature positions together to learn the attention with a shallow convolutional model.
- Semantics-Guided Neural Networks for Efficient Skeleton-Based Human Action Recognition645
A simple yet effective semantics-guided neural network (SGN) for skeleton-based action recognition that achieves the state-of-the-art performance on the NTU60, NTU120, and SYSU datasets.
- View Adaptive Neural Networks for High Performance Skeleton-Based Human Action Recognition522
A novel view adaptation scheme, which automatically determines the virtual observation viewpoints over the course of an action in a learning based data driven manner, and a two-stream scheme (referred to as VA-fusion) that fuses the scores of the two networks to provide the final prediction, obtaining enhanced performance.
- 496
- Style Normalization and Restitution for Generalizable Person Re-Identification425
The aim of this paper is to design a generalizable person ReID framework which trains a model on source domains yet is able to generalize/perform well on target domains, and to enforce a dual causal loss constraint in SNR to encourage the separation of identity-relevant features and identity-irrelevant features.
- 422
- PHASEN: A Phase-and-Harmonics-Aware Speech Enhancement Network412
This paper proposes a phase-and-harmonics-aware deep neural network (DNN), named PHASEN, which has the ability to handle detailed phase patterns and to utilize harmonic patterns, and outperforms previous methods by a large margin on four metrics.
- Efficient frequency domain selective scrambling of digital video391
A joint encryption and compression framework in which video data are scrambled efficiently in the frequency domain by employing selective bit scrambling, block shuffling and block rotation of the transform coefficients and motion vectors is presented.
- Inpaint Anything: Segment Anything Meets Image Inpainting334
The core idea behind IA is to combine the strengths of different models in order to build a very powerful and user-friendly pipeline for solving inpainting-related problems.
- 308
- Densely Semantically Aligned Person Re-Identification302
This work is the first to make use of fine grained semantics for addressing misalignment problems for re-ID and construct a set of densely semantically aligned part images (DSAP-images), where the same spatial positions have the same semantics across different person images.
- SPM-Tracker: Series-Parallel Matching for Real-Time Visual Object Tracking263
The greatest challenge facing visual object tracking is the simultaneous requirements on robustness and discrimination power, and a SiamFC-based tracker, named SPM-Tracker, is proposed, to address the two requirements in two separate matching stages.
- 257
- Cross View Fusion for 3D Human Pose Estimation255
This work introduces a cross-view fusion scheme into CNN to jointly estimate 2D poses for multiple views and presents a recursive Pictorial Structure Model to recover the 3D pose from the multi-view 2D pose.
- 248
- VoxelPose: Towards Multi-camera 3D Human Pose Estimation in Wild Environment233
An end-to-end solution which directly operates in the $3$D space, therefore avoids making incorrect decisions in the 2D space and outperforms the state-of-the-arts on the public datasets.
- Online Human Action Detection Using Joint Classification-Regression Recurrent Neural Networks226
This paper proposes a multi-task end-to-end Joint Classification-Regression Recurrent Neural Network to better explore the action type and temporal localization information and demonstrates very promising performance of this scheme.
- A format-compliant configurable encryption framework for access control of video218
New methods of performing selective encryption and spatial/frequency shuffling of compressed digital content that maintain syntax compliance after content has been secured are introduced.
- A statistical watermark detection technique without using original images for resolving rightful ownerships of digital images212
It is shown that, for the particular application of resolving rightful ownership using invisible watermarks, it might be crucial to require that the original image not be directly involved in the watermark detection process.
- ReSTR: Convolution-free Referring Image Segmentation Using Transformers210
The first convolution-free model for referring image segmentation using transformers, dubbed ReSTR, is presented, which can capture long-range dependencies between entities within each modality and outperforms all existing models.
- Correlation-Aware Deep Tracking200
This work proposes a novel target-dependent feature network inspired by the self-/cross-attention scheme that deeply embeds cross-image feature correlation in multiple layers of the feature network and is able to suppress non-target features, resulting in instance-varying feature extraction.
- Exploiting Sample Uncertainty for Domain Adaptive Person Re-Identification194
This paper proposes to estimate and exploit the credibility of the assigned pseudo-label of each sample to alleviate the influence of noisy labels, by suppressing the contribution of noisy samples, and introduces uncertainty-guided optimization, which brings significant improvement and achieves the state-of-the-art performance on benchmark datasets.
- Tracking by Instance Detection: A Meta-Learning Approach193
This work proposes a principled three-step approach to build a high-performance tracker based on two modern detectors RetinaNet and FCOS, and finds that model-agnostic meta-learning (MAML) offers a strategy to initialize the detector that satisfies the authors' needs.
- STOmicsDB: a comprehensive database for spatial transcriptomics data sharing, analysis and visualization184
The Spatial TranscriptOmics DataBase (STOmicsDB) is presented, a database that serves as a one-stop hub for spatial transcriptomics and offers a distinctive capability of customizing dedicated sub-databases in STOmicsDB for researchers, assisting them in visualizing their spatial transcriptomic analyses.
- 182
- 169
- Semantics-Aligned Representation Learning for Person Re-Identification164
A framework that drives the reID network to learn semantics-aligned feature representation through delicate supervision designs is proposed and achieves the state-of-the-art performances on the benchmark datasets CUHK03, Market1501, MSMT17, and the partial person reID dataset Partial REID.
- 136
- MetaAlign: Coordinating Domain Alignment and Classification for Unsupervised Domain Adaptation135
This paper proposes an effective meta-optimization based strategy dubbed MetaAlign, which treats the domain alignment objective and the classification objective as the meta-train and meta-test tasks in a meta-learning scheme, and encourages both tasks to be optimized in a coordinated way.
- Geometric-structure-based error concealment with novel applications in block-based low-bit-rate coding131
Experimental results show that the proposed coding and rate-shaping systems can provide significant subjective and objective gains over conventional approaches.
- 130
- Sparse MLP for Image Recognition: Is Self-Attention Really Necessary?128
An attention-free network called sMLPNet is built based on the existing MLP-based vision models, replacing the MLP module in the token-mixing step with a novel sparse MLP (sMLP) module that significantly reduces the number of model parameters and computational complexity.
- Multi-Granularity Reference-Aided Attentive Feature Aggregation for Video-Based Person Re-Identification128
This paper proposes an attentive feature aggregation module, namely Multi-Granularity Reference-aided Attentive Feature Aggregation (MG-RAFA), to delicately aggregate spatio-temporal features into a discriminative video-level feature representation.
- Uncertainty-Aware Multi-Shot Knowledge Distillation for Image-Based Object Re-Identification127
This paper designs an Uncertainty-aware Multi-shot Teacher-Student (UMTS) Network that consists of a teacher network that learns the comprehensive features from multiple images of the same object, and a student network that takes a single image as input.
- AdaFuse: Adaptive Multiview Fusion for Accurate Human Pose Estimation in the Wild127
AdaFuse is an adaptive multiview fusion method, which can enhance the features in occluded views by leveraging those in visible views by exploring the sparsity of the heatmap representation.
- A Simple Baseline for Multi-Object Tracking.122
This work studies the essential reasons behind the failure of object detection and re-identification, and presents a simple baseline that remarkably outperforms the state-of-the-arts on the public datasets at $30$ fps.
- Style Normalization and Restitution for Domain Generalization and Adaptation113
A novel Style Normalization and Restitution module (SNR) is designed to simultaneously ensure high generalization and discrimination capability of the networks and enforce a dual restitution loss constraint to encourage the better separation of task-relevant and task-irrelevant features.
- Object Detection in Videos by High Quality Object Linking113
This paper proposes a short tubelet linking algorithm that links temporally-overlapping short tubelets to form long tubelets and improves results by 8.8 percent over the static image detector for fast moving objects.
- 113
- ActFormer: A GAN-based Transformer towards General Action-Conditioned 3D Human Motion Generation110
A GAN-based Transformer for general action-conditioned 3D human motion generation, including not only single-person actions but also multi-person interactive actions, and introduces a new synthetic dataset of complex multi-person combat behaviors.
- RESIDE: A Benchmark for Single Image Dehazing.110
This paper presents a new large-scale benchmark consisting of both synthetic and real-world hazy images, called REalistic Single Image DEhazing (RESIDE), and provides a rich variety of criteria for dehazing algorithm evaluation, ranging from full-reference metrics, to no-reference metric, to subjective evaluation and the novel task-driven evaluation.
- 105
- When Shift Operation Meets Vision Transformer: An Extremely Simple Alternative to Attention Mechanism104
The role of attention mechanism is simplified into an extremely simple case: ZERO FLOP and ZERO parameter, and a new backbone network is constructed, namely ShiftViT, where the attention layers in ViT are substituted by shift operations.
- Adding Attentiveness to the Neurons in Recurrent Neural Networks102
The effectiveness of the proposed EleAtt-RNN is demonstrated by applying it to the action recognition tasks on both 3D human skeleton data and RGB videos.
- 102
- 97
- VoxelTrack: Multi-Person 3D Human Pose Estimation and Tracking in the Wild95
VoxelTrack employs a multi-branch network to jointly estimate 3D poses and re-identification (Re-ID) features for all people in the environment and outperforms the state-of-the-art methods by a large margin on four public datasets.
- SocialTransfer94
This paper investigates how to build a mutual connection among the disparate social media on the Internet, using which cross- domain media recommendation can be realized through SocialTransfer, a novel cross-domain real-time transfer learning framework.
- An overview of the visual optimization tools in JPEG 200091
It is shown that the visual tool sets in JPEG 2000 are much richer than what is achievable in JPEG, where only spatially invariant frequency weighting can be exploited.
- Pose-Guided Feature Learning with Knowledge Distillation for Occluded Person Re-Identification87
A network named Pose-Guided Feature Learning with Knowledge Distillation (PGFL-KD), where the pose information is exploited to regularize the learning of semantics aligned features but is discarded in testing to achieve high accuracy while preserving low inference complexity.
- Joint Time-Frequency and Time Domain Learning for Speech Enhancement86
A cross-domain framework named TFT-Net, which takes time-frequency spectrogram as input and produces time-domain waveform as output is presented, which achieves the highest SDR and SSNR among state-of-the-art methods on two major speech enhancement benchmarks.
- Fusing Wearable IMUs With Multi-View Images for Human Pose Estimation: A Geometric Approach83
A geometric approach is presented to reinforce the visual features of each pair of joints based on the IMUs to improve 2D pose estimation accuracy especially when one joint is occluded.
- Efficient frequency domain video scrambling for content access control83
A joint encryption and compression framework in which video data are scrambled efficiently in the frequency domain by employing selective bit scrambling, block shuffling and block rotation of the transform coefficients and motion vectors is presented.
- Moving Indoor: Unsupervised Video Depth Learning in Challenging Environments78
This work proposes a new optical-flow based training paradigm which reduces the difficulty of unsupervised learning by providing a clearer training target and handles the non-texture regions.
- Fast Discrete Cross-modal Hashing With Regressing From Semantic Labels78
A novel fast discrete cross-modal hashing method with regressing from semantic labels to take advantage of supervised labels to improve retrieval performance, which not only accelerates the hash learning process, but also helps generate stable hash codes.
- Unsupervised High-Resolution Depth Learning From Videos With Dual Networks77
This work proposes a simple yet effective dual networks architecture, which can directly take high- resolution images as input and generate high-resolution and high-accuracy depth map efficiently and proposes a Self-assembled Attention (SA-Attention) module to handle low-texture region.
- Learning Tracking Representations via Dual-Branch Fully Transformer Networks76
A Siamese-like Dual-branch network based on solely Transformers for tracking, which outperforms the state-of-the-art methods on the GOT-10k and VOT2020 benchmarks and achieves real-time inference speed (about 40 fps) on one GPU.
- Multi-Grained Attention Networks for Single Image Super-Resolution76
This paper proposes a unified approach called “multi-grained attention networks (MGAN)” which fully exploits the advantages of multi-scale and attention mechanisms in SR tasks and shows the superiority in terms of both accuracy and model size.
- Global Distance-Distributions Separation for Unsupervised Person Re-identification75
A global distance-distributions separation (GDS) constraint over the two distributions is introduced to encourage the clear separation of positive and negative samples from a global view and validation of the effectiveness of the GDS constraint in unsupervised ReID networks is validated.
- Detect or Track: Towards Cost-Effective Video Object Detection/Tracking73
A scheduler network, which determines to detect or track at a certain frame, as a generalization of Siamese trackers is proposed, which is more effective than the frame skipping baselines and flow-based approaches in video object detection/tracking.
- Human Pose Estimation Using Global and Local Normalization71
A two-stage normalization scheme, human body normalization and limb normalization, is presented to make the distribution of the relative joint locations compact, resulting in easier learning of convolutional spatial models and more accurate pose estimation.
- Beyond Intra-modality: A Survey of Heterogeneous Person Re-identification69
This paper provides a comprehensive review of state-of-the-art Hetero-ReID methods that address the challenge of inter-modality discrepancies and compares the representative approaches from two perspectives, i.e., the application scenario and the learning pipeline.
- 66
- 66
- 66
- Temporal–Spatial Mapping for Action Recognition64
This work introduces a simple yet effective operation, termed temporal–spatial mapping, for capturing the temporal evolution of the frames by jointly analyzing all the frames of a video, and proposes a temporal attention model within a shallow convolutional neural network to efficiently exploit the temporal-spatial dynamics.
- 63
- Digital image watermarking using visual models63
A watermarking technique for digital images that is based on utilizing visual models which have been developed in the context of image compression that is shown to provide dramatic improvement over the current state-of-the-art both in terms of transparency and robustness.
- View Invariant 3D Human Pose Estimation62
A view-invariant 3D human pose estimation module to alleviate the effects of viewpoint diversity and shows robustness for different 3D pose base networks and achieves a significant improvement over a powerful baseline on the public3D pose estimation benchmark Human3.6M.
- 62
- S2R-DepthNet: Learning a Generalizable Depth-specific Structural Representation61
This work is the first to explore the learning of a depth-specific structural representation, which captures the essential feature for depth estimation and ignores irrelevant style information.
- Towards a Better Match in Siamese Network Based Visual Object Tracker59
This paper proposes two simple yet effective mechanisms, namely angle estimation and spatial masking, to address issues of Siamese network based trackers, and produces a tracker, named Siam-BM, which is the best-performing realtime tracker to date.
- Self-Supervised Visual Representations Learning by Contrastive Mask Prediction57
A novel contrastive mask prediction (CMP) task for visual representation learning and a mask contrast (MaskCo) framework to implement the idea, which provides a promising alternative to the ID-based methods for self-supervised learning in the wild.
- 57
- 57
- 56
- 55
- Scalable streaming of JPEG2000 images using hypertext transfer protocol55
This paper describes a scalable architecture for streaming of JPEG2000 images, using Hypertext Transfer Protocol (HTTP), which will ease its deployment, because of the widespread availability and accessibility of web servers.
- 54
- A CNN-Based Approach for Automatic License Plate Recognition in the Wild54
This paper addresses automatic license plate recognition (ALPR) in the wild with a cascade structure comprising of a fast region proposal network and a R-CNN network, and proposes an innovative structure composed of parallel spatial transform networks and shared-weight recognizers.
- Lifelong Unsupervised Domain Adaptive Person Re-identification with Coordinated Anti-forgetting and Adaptation53
An effective scheme for this task, dubbed CLUDA-ReID, where the anti-forgetting is harmoniously coordinated with the adaptation, and a meta-based Coordinated Data Replay strategy is proposed to replay old data and update the network with a coordinated optimization direction for both adaptation and memorization.
- Secure and robust image hashing via compressive sensing53
A novel algorithm for generating a secure and robust image hash based onCompressive sensing and Fourier-Mellin transform and the property of dimension reduction inherent in compressive sensing for hash design is exploited.
- 50
- 50
- 49
- A Multi-band Wavelet Watermarking Scheme49
This paper presents a new multi-band wavelet watermarking scheme that can successfully resist common signal processing such as JPEG compression with quality factor as low as 15, and some geometric distortions such as cropping.
- 47
- Time delay and Doppler estimation for wideband acoustic signals in multipath environments47
A high-resolution method for jointly estimating the multipath time delays, Doppler scales, and attenuation amplitudes of a time-varying acoustical channel that can handle both narrowband and wideband signals is proposed.
- 47
- Fast estimation of sparse doubly spread acoustic channels46
A fast projected gradient method (FPGM) that can handle complex-valued data for estimating the delay-Doppler spread function of a time-varying channel is proposed and is robust to parameter selection and is orders-of-magnitude faster than standard convex optimization algorithms.
- AttributeNet: Attribute enhanced vehicle re-identification45
This work introduces AttributeNet (ANet) that jointly extracts identity-relevant features and attribute features and proposes a constraint, named Amelioration Constraint (AC), which encourages the feature after adding attribute features onto the general ReID feature to be more discriminative than the original general ReID feature.
- 44
- 44
- 44
- 44
- 43
- 43
- Skeleton-Based Mutually Assisted Interacted Object Localization and Human Action Recognition42
A joint learning framework for mutually assisted “interacted object localization” and “human action recognition” based on skeleton data is proposed and the temporal consistency of interacted object is explored as constraint to better localize the interacted object with the absence of ground-truth labels.
- 42
- Visual masking in wavelet compression for JPEG-200041
A nonuniform quantization scheme for JPEG2000 that leverages the masking properties of the visual system, in which visibility to distortions declines as image energy increases, and these techniques allow image quality increases in applications where using theVisual system's frequency response provides little advantage.
- Uncertainty-Aware Few-Shot Image Classification38
This work proposes Uncertainty-Aware Few-Shot framework for image classification by modeling uncertainty of the similarities of query-support pairs and performing uncertainty-aware optimization and achieves the state-of-the-art performance.
- 38
- An Empirical Study of the Collapsing Problem in Semi-Supervised 2D Human Pose Estimation37
This work presents a surprisingly simple approach to drive the model to learn in the correct direction and applies it to the state-of-the-art pose estimators and it further improves their performance on three public datasets.
- 36
- 35
- Predicting Future Instance Segmentation with Contextual Pyramid ConvLSTMs34
A novel framework is proposed, which contains a set of ConvLSTMs to exploit intra-level spatio-temporal contexts for predicting features of each individual level, and which can achieve state-of-the-art performance on two video instance segmentation benchmarks.
- Photo Stylistic Brush: Robust Style Transfer via Superpixel-Based Bipartite Graph34
This paper proposes a photo stylistic brush, an automatic robust style transfer approach based on Super pixel-based BIpartite Graph (SuperBIG), which guides SuperBIG to perform the transformation in a decorrelated color space.
- 34
- The China National GeneBank Sequence Archive (CNSA) 2024 update33
The state-of-the-art enhancements of CNSA are presented encompassing the dramatical increase of varied types of data, the latest features and services implemented in CNSA as well as consistent efforts supporting global cooperation in biodiversity preservation and utilization.
- Spatiotemporal Fusion in 3D CNNs: A Probabilistic View33
This paper proposes to convert the spatiotemporal fusion strategies into a probability space, which allows us to perform network-level evaluations of various fusion strategies without having to train them separately, and generates new fusion strategies which achieve the state-of-the-art performance on four well-known action recognition datasets.
- 31
- 31
- 29
- 29
- 29
- 29
- 28
- 28
- 27
- 27
- 26
- 26
- 26
- 25
- 25
- 25
- 23
- 23
- 23
- 23
- 22
- Re-energizing Domain Discriminator with Sample Relabeling for Adversarial Domain Adaptation22
This work proposes an efficient optimization strategy named Re-enforceable Adversarial Domain Adaptation (RADA), which aims to re-energize the domain discriminator during the training by using dynamic domain labels, and relabeling makes the less separable distributions more separable, and thus leads to a more powerful domain classifier w.r.t. the new data distributions.
- 22
- Beyond Triplet Loss: Meta Prototypical N-Tuple Loss for Person Re-identification20
A multi-class classification loss, i.e., N-tuple loss, is introduced to jointly consider multiple ($N$) instances for per-query optimization for person ReID, and achieves the state-of-the-art performance on the benchmark person Re ID datasets.
- 20
- 19
- 18
- A Survey of Behavior Foundation Model: Next-Generation Whole-Body Control System of Humanoid Robots17
This paper presents a comprehensive overview of BFMs for humanoid WBC, tracing their development across diverse pre-training pipelines, and discusses real-world applications, current limitations, urgent challenges, and future opportunities.
- 17
- 17
- 17
- 16
- 15
- 15
- 15
- 15
- 15
- 15
- 15
- 15
- One at a Time: Progressive Multi-Step Volumetric Probability Learning for Reliable 3D Scene Perception14
A multi-step learning framework dedicated to progressively refining the Volumetric Probability in a Diffusion process, dubbed as VPD, is introduced, which first build a coarse probability volume from input images with the off-the-shelf scene perception baselines, which is then conditioned as the basic geometry prior before being fed into a 3D diffusion UNet.
- 14
- 14
- 14
- 13
- 13
- 13
- Hierarchical Context Alignment With Disentangled Geometric and Temporal Modeling for Semantic Occupancy Prediction12
Hi-SOP first disentangles the geometric and temporal context for separate alignment, which two branches are then composed to enhance the reliability of SOP, which outperforms SOTAs for semantic scene completion on the SemanticKITTI & NuScenes-Occupancy datasets and LiDAR semantic segmentation on the NuScenes dataset.
- 12
- 12
- 12
- 12
- 12
- 12
- 12
- 11
- 11
- 11
- Relation-Aware Global Attention.11
This paper proposes an effective Relation-Aware Global Attention module for CNNs to fully exploit the global correlations to infer the attention and demonstrates the general applicability of RGA to vision tasks by applying it to the scene segmentation and image classification tasks resulting in consistent performance improvement.
- 11
- 11
- 11
- 11
- 10
- 10
- 10
- 10
- 10
- 10
- 10
- 9
- 9
- 9
- 9
- 9
- 9
- 9
- 9
- 8
- 8
- General-Purpose Speech Representation Learning through a Self-Supervised Multi-Granularity Framework8
- 8
- 8
- 8
- 8
- 8
- 8
- 8
- 8
- 8
- 8
- 8
- 7
- 7
- 7
- 7
- 7
- 7
- CASINet: Content-Adaptive Scale Interaction Networks for scene parsing7
A Content-Adaptive Scale Interaction Network (CASINet) to exploit the multi-scale features for scene parsing based on the classic Atrous Spatial Pyramid Pooling module, followed by a proposed contextual scale interaction (CSI) module, and a scale adaptation (SA) module.
- 7
- 7
- 7
- 7
- 7
- 7
- 7
- 7
- Knowledge Regularized Negative Feature Tuning of Vision-Language Models for Out-of-Distribution Detection6
This work proposes a novel method called Knowledge Regularized Negative Feature Tuning (KR-NFT), which integrates an innovative adaptation architecture termed Negative Feature Tuning (NFT) and a corresponding knowledge-regularization (KR) optimization strategy, designed to enhance the discrimination between ID and OOD sets while mitigating pre-trained knowledge forgetting.
- 6
- 6
- 6
- 6
- 6
- 6
- 6
- 6
- 6
- 6
- 5
- 5
- 5
- Rethinking Classification Loss Designs for Person Re-identification with a Unified View.5
It is argued that triplet-based optimization can be viewed as a two-class subsampling classification, which performs classification over two sampled categories based on instance similarities, and a new meta prototypical N-tuple loss is proposed for more efficient multi- class subsampled classification.
- 5
- 5
- 5
- 5
- 5
- 5
- 5
- 5
- 5
- 5
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- 4
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- 3
- Novel approach to unequal error protection for image/video delivery with source detector3
It is shown that by intelligently combining unequal channel protection and source detection, both subjectively and objectively better quality of reconstructed images can be obtained.
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- NaviNeRF++: Towards Interpretable 3D Reconstruction via Unsupervised Disentangled Representation Learning1
This work achieves interpretable 3D reconstruction and fine-grained 3D disentanglement in an unsupervised manner, and is the first to achieve interpretable 3D reconstruction and fine-grained 3D disentanglement in an unsupervised manner.
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- 1
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
- –
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 2026-10-13. 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.