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
- Shaogang GongSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works328 from public data
- Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective with Transformers3,852
This paper deploys a pure transformer to encode an image as a sequence of patches, termed SEgmentation TRansformer (SETR), and shows that SETR achieves new state of the art on ADE20K, Pascal Context, and competitive results on Cityscapes.
- Pixel2Mesh: Generating 3D Mesh Models from Single RGB Images1,529
An end-to-end deep learning architecture that produces a 3D shape in triangular mesh from a single color image by progressively deforming an ellipsoid, leveraging perceptual features extracted from the input image.
- Soft Filter Pruning for Accelerating Deep Convolutional Neural Networks1,096
The proposed Soft Filter Pruning (SFP) method enables the pruned filters to be updated when training the model after pruning, which has two advantages over previous works: larger model capacity and less dependence on the pretrained model.
- Transductive Multi-View Zero-Shot Learning548
A novel heterogeneous multi-view hypergraph label propagation method is formulated for zero-shot learning in the transductive embedding space that rectifies the projection shift between the auxiliary and target domains, exploits the complementarity of multiple semantic representations, and significantly outperforms existing methods for both zero- shot and N-shot recognition.
- Pose-Normalized Image Generation for Person Re-identification483
This work proposes a novel deep person image generation model for synthesizing realistic person images conditional on the pose based on a generative adversarial network designed specifically for pose normalization in re-id, thus termed pose-normalization GAN (PN-GAN).
- Chained-Tracker: Chaining Paired Attentive Regression Results for End-to-End Joint Multiple-Object Detection and Tracking382
The two major novelties: chained structure and paired attentive regression, make CTracker simple, fast and effective, setting new MOTA records on MOT16 and MOT17 challenge datasets (67.6 and 66.6, respectively), without relying on any extra training data.
- Learning Salient Boundary Feature for Anchor-free Temporal Action Localization329
This paper proposes the first purely anchor-free temporal localization method, which is both efficient and effective, and beats all anchor-based and actionness-guided methods with a remarkable margin on THUMOS14, achieving state-of-the-art results, and comparable ones on ActivityNet v1.3.
- Multi-scale Deep Learning Architectures for Person Re-identification311
A novel multi-scale deep learning model is proposed that is able to learn deep discriminative feature representations at different scales and automatically determine the most suitable scales for matching.
- Pixel2Mesh++: Multi-View 3D Mesh Generation via Deformation291
This model learns to predict series of deformations to improve a coarse shape iteratively and exhibits generalization capability across different semantic categories, number of input images, and quality of mesh initialization.
- 259
- Image Deformation Meta-Networks for One-Shot Learning255
This work combines a meta-learner with an image deformation sub-network that produces additional training examples, and optimize both models in an end-to-end manner to significantly outperform state-of-the-art approaches on widely used one-shot learning benchmarks.
- Transductive Multi-view Embedding for Zero-Shot Recognition and Annotation251
This paper proposes a novel framework, transductive multi-view embedding, that rectifies the projection shift between the auxiliary and target domains, exploits the complementarity of multiple semantic representations, achieves state-of-the-art recognition results on image and video benchmark datasets, and enables novel cross-view annotation tasks.
- Long-Term Cloth-Changing Person Re-identification248
This work introduces a shape embedding module as well as a cloth-elimination shape-distillation module aiming to eliminate the now unreliable clothing appearance features and focus on the body shape information.
- Incremental Transformer Structure Enhanced Image Inpainting with Masking Positional Encoding220
The proposed model restores holistic image structures with a powerful attention-based transformer model in a fixed low-resolution sketch space and can be integrated with other pretrained inpainting models efficiently with the zero-initialized residual addition.
- Learning a Few-shot Embedding Model with Contrastive Learning207
This paper proposes a novel contrastive training scheme dubbed infoPatch, exploiting the patch-wise relationship to substantially improve the popular infoNCE and investigates the contrastive learning with Noise Contrastive Estimation in a supervised manner for training a few-shot embedding model.
- Instance Credibility Inference for Few-Shot Learning204
This paper presents a simple statistical approach, dubbed Instance Credibility Inference (ICI), to exploit the distribution support of unlabeled instances for few-shot learning to establish new state-of-the-arts on four widely used few- shot learning benchmark datasets.
- 195
- Asymptotic Soft Filter Pruning for Deep Convolutional Neural Networks192
An asymptotic soft filter pruning (ASFP) method to accelerate the inference procedure of the deep neural networks and shows the effectiveness of the ASFP on image classification benchmarks.
- Recent Advances in Zero-Shot Recognition: Toward Data-Efficient Understanding of Visual Content186
This article provides a comprehensive review of existing zero-shot recognition techniques covering various aspects ranging from representations of models, data sets, and evaluation settings and highlights the limitations of existing approaches.
- Learning to Score Figure Skating Sport Videos181
A deep architecture that includes two complementary components, i.e., Self-Attentive L STM and Multi-scale Convolutional Skip LSTM can efficiently learn the local and global sequential information in each video.
- Depth-conditioned Dynamic Message Propagation for Monocular 3D Object Detection170
A depth-conditioned dynamic message propagation (DDMP) network to effectively integrate the multi-scale depth information with the image context and dynamically predicting hybrid depth-dependent filter weights and affinity matrices for propagating information is proposed.
- When Person Re-identification Meets Changing Clothes167
It is found that changing clothes makes ReID a much harder problem in the sense of bringing difficulties to learning effective representations and also challenges the generalization ability of previous ReID models to identify persons with unseen (new) clothes.
- 148
- DeepSFM: Structure from Motion via Deep Bundle Adjustment142
This work designs a physical driven architecture, namely DeepSFM, inspired by traditional Bundle Adjustment, which consists of two cost volume based architectures for depth and pose estimation respectively, iteratively running to improve both.
- 136
- Semi-supervised Vocabulary-Informed Learning135
A maximum margin framework for semantic manifold-based recognition that incorporates distance constraints from (both supervised and unsupervised) vocabulary atoms, ensuring that labeled samples are projected closest to their correct prototypes, in the embedding space, than to others.
- Learning Dynamic Alignment via Meta-filter for Few-shot Learning134
This paper proposes to learn a dynamic alignment, which can effectively highlight both query regions and channels according to different local support information and facilitates dynamical knowledge adaptation for few-shot learning.
- 129
- Depth Guided Adaptive Meta-Fusion Network for Few-shot Video Recognition124
A depth guided Adaptive Meta-Fusion Network for few-shot video recognition which is termed as AMeFu-Net, which is trained in the meta-learning way and demonstrates the effectiveness of the model on several action recognition benchmarks.
- Attribute Learning for Understanding Unstructured Social Activity119
This paper contributes an unstructured social activity attribute (USAA) dataset, introduces the concept of semi-latent attribute space, and proposes a novel model for learning the latent attributes which alleviate the dependence of existing models on exact and exhaustive manual specification of the attribute-space.
- Image Block Augmentation for One-Shot Learning114
This paper proposes a Self-Training Jigsaw Augmentation (Self-Jig) method for one-shot learning, which provides a very useful way of directly generating massive amounts of training images for novel classes.
- SAR-Net: Shape Alignment and Recovery Network for Category-level 6D Object Pose and Size Estimation111
This paper proposes a method of Category-level 6D Object Pose and Size Estimation (COPSE) from the point cloud of the target object, without external real pose-annotated training data, and deploys the approach to a physical Baxter robot to perform grasping tasks on unseen but category-known instances.
- StyleAdv: Meta Style Adversarial Training for Cross-Domain Few-Shot Learning110
A novel model-agnostic meta Style Adversarial training (StyleAdv) method together with a novel style adversarial attack method is proposed for CD-FSL, which boosts the generalization ability for novel target datasets.
- Grad-PU: Arbitrary-Scale Point Cloud Upsampling via Gradient Descent with Learned Distance Functions108
The method first interpolates the low-res point cloud according to a given upsampling rate, and then refine the positions of the interpolated points with an iterative optimization process, guided by a trained model estimating the difference between the current point cloud and the high-res target.
- MemFlow: Optical Flow Estimation and Prediction with Memory102
This work presents MemFlow, a real-time method for optical flow estimation and prediction with memory that outperforms VideoFlow with fewer parameters and faster inference speed on Sintel and KITTI-15 datasets in terms of generalization performance.
- Generalized Meta-FDMixup: Cross-Domain Few-Shot Learning Guided by Labeled Target Data101
A novel Generalized Meta-learning based Feature-Disentangled Mixup network, namely GMeta-FDMixup is proposed to address Cross-Domain Few-Shot Learning (CD-FSL), and a novel feature disentanglement module is contributed to narrow the domain gap explicitly.
- 101
- ONCE-3DLanes: Building Monocular 3D Lane Detection98
This work presents ONCE-3DLanes, a real-world autonomous driving dataset with lane layout annotation in 3D space, and presents an extrinsic-free, anchorfree method, called SALAD, regressing the 3D coordinates of lanes in image view without converting the feature map into the bird's-eye view (BEV).
- 98
- 97
- Stacked Semantics-Guided Attention Model for Fine-Grained Zero-Shot Learning97
A novel stacked semantics-guided attention (S2GA) model to obtain semantic relevant features by using individual class semantic features to progressively guide the visual features to generate an attention map for weighting the importance of different local regions.
- Density-preserving Deep Point Cloud Compression93
A novel deep point cloud compression method that preserves local density information by encoding local geometry and density with three embeddings: density embedding, local position embedding and ancestor embedding.
- 93
- Interestingness Prediction by Robust Learning to Rank91
A more principled way to identify annotation outliers is proposed by formulating theinterestingness prediction task as a unified robust learning to rank problem, tackling both the outlier detection and interestingness prediction tasks jointly.
- 89
- MVSFormer++: Revealing the Devil in Transformer's Details for Multi-View Stereo88
This paper introduces MVSFormer++, a method that prudently maximizes the inherent characteristics of attention to enhance various components of the MVS pipeline, and achieves state-of-the-art performance on the challenging DTU and Tanks-and-Temples benchmarks.
- InternVLA-M1: A Spatially Guided Vision-Language-Action Framework for Generalist Robot Policy85
InternVLA-M1 outperforms its variant without spatial guidance in long-horizon reasoning-intensive scenarios, and surpassed existing works by over 10%.
- Sketch-BERT: Learning Sketch Bidirectional Encoder Representation From Transformers by Self-Supervised Learning of Sketch Gestalt85
This work presents a model of learning Sketch Bidirectional Encoder Representation from Transformer (Sketch-BERT), and generalizes BERT to sketch domain, with the novel proposed components and pre-training algorithms, including the newly designed sketch embedding networks, and the self-supervised learning of sketch gestalt.
- 84
- Adaptively Weighted Multi-task Deep Network for Person Attribute Classification75
A novel validation loss trend algorithm is proposed to dynamically and adaptively update the weight for learning each task in the training process to solve the problems of multi-task deep convolutional neural network for person attribute analysis.
- ZITS++: Image Inpainting by Improving the Incremental Transformer on Structural Priors74
This study study learning a Zero-initialized residual addition based Incremental Transformer on Structural priors (ZITS++), an improved model upon the authors' conference work, ZITS (Dong et al. 2022).
- Delving into Data: Effectively Substitute Training for Black-box Attack73
A novel perspective substitute training that focuses on designing the distribution of data used in the knowledge stealing process and the combination of these two modules can further boost the consistency of the substitute model and target model, which greatly improves the effectiveness of adversarial attack.
- M$^3$Lung-Sys: A Deep Learning System for Multi-Class Lung Pneumonia Screening From CT Imaging71
The quantitative results with plenty of metrics indicate the superiority of the proposed model on both slice- and patient-level classification tasks, and the generated lesion location maps make the system interpretable and more valuable to clinicians.
- Ranking Distance Calibration for Cross-Domain Few-Shot Learning70
This work starts from a generic representation pre-trained by a cross-entropy loss and a conventional distance-based classifier, along with an image retrieval view, to employ a re-ranking process to calibrate a target distance matrix by discovering the k-reciprocal neighbours within the task.
- Rethinking Optical Flow from Geometric Matching Consistent Perspective68
The proposed MatchFlow model employs a QuadTree attention-based network pre-trained on MegaDepth to extract coarse features for further flow regression, and has great cross-dataset generalization.
- Learning a Sketch Tensor Space for Image Inpainting of Man-made Scenes68
This paper proposes a Multi-scale Sketch Tensor inpainting (MST) network, with a novel encoder-decoder structure, that can achieve competitive performance in inPainting general nature images over the competitors.
- Neural Pose Transfer by Spatially Adaptive Instance Normalization67
This paper proposes the first neural pose transfer model that solves the pose transfer via the latest technique for image style transfer, leveraging the newly proposed component -- spatially adaptive instance normalization.
- Multi-task Deep Neural Network for Joint Face Recognition and Facial Attribute Prediction65
This paper for the first time, advocate a multi-task deep neural network for jointly learning face recognition and facial attribute prediction tasks and can achieve the best performance on the small protocol among all the publicly released methods.
- Semantic Feature Augmentation in Few-shot Learning.60
A novel auto-encoder network dual TriNet, which projects multi-layer visual features of deep CNNs into the semantic space, and the augmented instance representation is projected back into the image feature spaces by the decoder TriNet.
- Robust Subjective Visual Property Prediction from Crowdsourced Pairwise Labels58
A more principled way to identify annotation outliers is proposed by formulating the subjective visual property prediction task as a unified robust learning toRank problem, tackling both the outlier detection and learning to rank jointly.
- Video Emotion Recognition with Transferred Deep Feature Encodings57
A novel auxiliary Image Transfer Encoding process is proposed to efficiently encode and generate video representation and thoroughly investigate different configurations of convolutional neural networks.
- Transductive Multi-label Zero-shot Learning56
This paper proposes a multi-output deep regression model to project an image into a semantic word space, which explicitly exploits the correlations in the intermediate semantic layer of word vectors, and a novel zero-shot learning algorithm for multi-label data that exploits the unique compositionality property of semantic word vector representations.
- How to Trust Unlabeled Data? Instance Credibility Inference for Few-Shot Learning55
A statistical approach, dubbed Instance Credibility Inference (ICI), to exploit the support of unlabeled instances for few-shot visual recognition to collect all the correctly-predicted pseudo-labeled instances from the noisy pseudo- labeled set.
- A Large-Scale Attribute Dataset for Zero-Shot Learning55
A Large-scale Attribute Dataset with 78,017 images of 230 classes with 359 attributes of visual, semantic and subjective properties is proposed, which argues that the "co-occurrence bias problem" of existing datasets, which is caused by the biased co-occurring of objects, significantly hinders models from correctly learning the concept.
- TC-Net for iSBIR55
A Triplet Classification Network (TC-Net) for iSBIR is presented which is composed of two major components: triplet Siamese network, and auxiliary classification loss which can break the limitations existed in previous works.
- Harnessing Synthesized Abstraction Images to Improve Facial Attribute Recognition52
This paper builds a dual-path facial attribute recognition network to utilize features from the original face images and facial abstraction images to improve facial attributes recognition, especially the attributes located on small face regions.
- Coarse-to-Fine Amodal Segmentation with Shape Prior50
A novel approach, called Coarse-to-Fine Segmentation (C2F-Seg), that addresses this problem by progressively modeling the amodal segmentation by introducing a convolution refine module to inject fine-grained information and provide a more precise amodal object segmentation based on visual features and coarse-predicted segmentation.
- Exploring Fine-Grained Representation and Recomposition for Cloth-Changing Person Re-Identification47
A novel FIne-grained Representation and Recomposition (FIRe2) framework to tackle both limitations without any auxiliary annotation or data is proposed and can achieve state-of-the-art performance on five widely-used cloth-changing person Re-ID benchmarks.
- Embodied One-Shot Video Recognition45
A novel setting is introduced, termed as embodied agents based one- shot learning, which leverages synthetic videos produced in a virtual environment to understand realistic videos of target classes and proposes a general video segment augmentation method, which significantly facilitates a variety of one-shot learning tasks.
- Learning to Generate Posters of Scientific Papers by Probabilistic Graphical Models45
A data-driven framework, which utilizes graphical models, is proposed, and the key elements of a good poster, including attributes of each panel and arrangements of graphical elements, are learned and inferred from data.
- NTIRE 2025 Challenge on Cross-Domain Few-Shot Object Detection: Methods and Results44
An overview of the 1st NTIRE 2025 CD-FSOD Challenge is presented, highlighting the proposed solutions and summarizing the results submitted by the participants.
- Recent Few-shot Object Detection Algorithms: A Survey with Performance Comparison43
An overview of FSOD, including the problem definition, common datasets, and evaluation protocols, is given and a taxonomy is proposed that groups FSOD methods into three types, i.e., data-oriented, model- oriented, and algorithm-oriented.
- 41
- 40
- Pixel2Mesh++: 3D Mesh Generation and Refinement From Multi-View Images40
This work adopts to further improve the shape quality by leveraging cross-view information with a graph convolution network, and learns to predict series of deformations to improve a coarse shape iteratively.
- Deep Ranking for Image Zero-Shot Multi-Label Classification40
This work investigates and formalizes a flexible framework consisting of two components, i.e., visual-semantic embedding and zero-shot multi-label prediction, and presents a deep regression model to project the visual features into the semantic space, which explicitly exploits the correlations in the intermediate semantic layer of word vectors and makes label prediction possible.
- Wave-SAN: Wavelet based Style Augmentation Network for Cross-Domain Few-Shot Learning39
This paper studies the problem of CD-FSL by spanning the style distributions of the source dataset, and proposes a novel Style Augmentation (StyleAug) module and a Self-Supervised Learning (SSL) module to ensure the predictions of style-augmented images are semantically similar to the unchanged ones.
- 39
- Adaptive Slot Attention: Object Discovery with Dynamic Slot Number38
An adaptive slot attention (AdaSlot) mecha-nism that dynamically determines the optimal number of slots based on the content of the data is introduced that exhibits the capability to dynamically adapt the slot number according to each instance's complexity.
- MVGenMaster: Scaling Multi-View Generation from Any Image via 3D Priors Enhanced Diffusion Model37
This work introduces MVGenMaster, a multi-view diffusion model enhanced with 3D priors to address versatile Novel View Synthesis (NVS) tasks, and presents several training and model modifications to strengthen the model with scaled-up datasets.
- GeoVLN: Learning Geometry-Enhanced Visual Representation with Slot Attention for Vision-and-Language Navigation37
GeoVLN is proposed, which learns Geometry-enhanced visual representation based on slot attention for robust Visual-and-Language Navigation and employs V&L BERT to learn a cross-modal representation that incorporate both language and vision informations.
- Scalable Penalized Regression for Noise Detection in Learning with Noisy Labels37
This paper proposes using a theoretically guaranteed noisy label detection framework to detect and remove noisy data for Learning with Noisy Labels (LNL), and designs a penalized regression to model the linear relation between network features and one-hot labels.
- Learning Compositional Representation for 4D Captures with Neural ODE37
A compositional representation for 4D captures, i.e. a deforming 3D object over a temporal span, that disentangles shape, initial state, and motion respectively is introduced and an Identity Exchange Training (IET) strategy is proposed to encourage the network to learn effectively decoupling each component.
- MVSFormer: Multi-View Stereo by Learning Robust Image Features and Temperature-based Depth36
A pre-trained ViT enhanced MVS network called MVSFormer, which can learn more reliable feature representations benefited by informative priors from ViT, and achieves state-of-the-art performance on the DTU dataset.
- A Multi-Task Neural Approach for Emotion Attribution, Classification, and Summarization36
A new neural approach is proposed, Bi-stream Emotion Attribution-Classification Network (BEAC-Net), to solve three related emotion analysis tasks: emotion recognition, emotion attribution, and emotion-oriented summarization, in a single integrated framework.
- 35
- A Unified Efficient Pyramid Transformer for Semantic Segmentation35
This work proposes a unified framework (UN-EPT) to segment objects by considering both context information and boundary artifacts, and demonstrates promising performance on three popular benchmarks for semantic segmentation with low memory footprint.
- 35
- VividPose: Advancing Stable Video Diffusion for Realistic Human Image Animation34
This work proposes VividPose, an innovative end-to-end pipeline based on Stable Video Diffusion (SVD) that ensures superior temporal stability and enhances the retention of human identity, and introduces a geometry-aware pose controller that utilizes both dense rendering maps from SMPL-X and sparse skeleton maps.
- 34
- PatchMix Augmentation to Identify Causal Features in Few-Shot Learning33
A novel data augmentation strategy dubbed as PatchMix that can break this spurious dependency between patch-level information and supervision of the query images with random gallery images from different classes from the query ones is proposed and theoretically shows that such an augmentation mechanism, different from existing ones, is able to identify the causal features.
- 33
- Learning to Generate Posters of Scientific Papers33
A data-driven framework, that utilizes graphical models, is proposed, that learns and inferred the key elements of a good poster, including panel layout and attributes of each panel, are learned and inferred from data.
- 32
- Towards Enhanced Image Inpainting: Mitigating Unwanted Object Insertion and Preserving Color Consistency31
A post-processing approach, dubbed as ASUKA (Aligned Stable inpainting with UnKnown Areas prior), to improve inpainting models and mitigates object hallucination and improves color consistency over standard diffusion and rectified flow models and other inpainting methods.
- ReasonGrounder: LVLM-Guided Hierarchical Feature Splatting for Open-Vocabulary 3D Visual Grounding and Reasoning31
ReasonGrounder is proposed, an LVLM-guided framework that uses hierarchical 3D feature Gaussian fields for adaptive grouping based on physical scale, enabling open-vocabulary 3D grounding and reasoning and significantly improves 3D grounding accuracy in real-world scenarios.
- Take Goods from Shelves: A Dataset for Class-Incremental Object Detection31
A new hierarchical large-scale object detection dataset, called Take Goods from Shelves (TGFS), containing 38K images of 24 fine-grained and 3 coarse classes is proposed and a preliminary method for solving the goods-adding problem, called Faster R-CNN Class-incremental Object Detector (FCIOD), is described and evaluated.
- Learning to Generate and Edit Hairstyles30
A new large-scale hairstyle dataset called Hairstyle30k is contributed, which is composed of 30k images containing 64 different types of hairstyles and a novel GAN model termed Hairstyles GAN (H-GAN) which can be learned efficiently is proposed.
- 29
- Vision Transformers: From Semantic Segmentation to Dense Prediction29
This work explores the global context learning potentials of ViTs for dense visual prediction and forms a family of Hierarchical Local-Global (HLG) Transformers, characterized by local attention within windows and global-attention across windows in a pyramidal architecture.
- LeftRefill: Filling Right Canvas based on Left Reference through Generalized Text-to-Image Diffusion Model29
LeftRefill is introduced, an innovative approach to efficiently harness large Text-to-Image (T2I) diffusion models for reference-guided image synthesis, based on the pre-trained StableDiffusion, to address two different challenges: reference-guided inpainting and novel view synthesis.
- A SLAM Algorithm Based on Adaptive Cubature Kalman Filter29
An innovative cubatureKalman filter-based SLAM (CKF-SLAM) algorithm based on an adaptive cubature Kalman filter (ACKF) was established and can reduce the state estimated error significantly and improve the navigation accuracy of the SLAM system effectively.
- VidCRAFT3: Camera, Object, and Lighting Control for Image-to-Video Generation28
VidCRAFT3 is presented, a unified and flexible I2V framework that explicitly models cross-factor interactions among geometry, motion, and illumination, enabling both independent and joint control over camera motion, object motion, and lighting direction.
- 28
- Improving Transformer-based Image Matching by Cascaded Capturing Spatially Informative Keypoints28
This work proposes a transformer-based cascade matching model, CasMTR, to efficiently learn dense feature correlations, which allows us to choose more reliable matching pairs for the relative pose estimation and achieves state-of-the-art performance in indoor and outdoor pose estimation as well as visual localization.
- Deep Hybrid Self-Prior for Full 3D Mesh Generation28
A novel hybrid 2D-3D self-prior in deep neural networks is proposed to significantly improve the geometry quality and produce a high-resolution texture map, which is typically missing from the output of commodity-level 3D scanners.
- Exploring Efficient Few-shot Adaptation for Vision Transformers27
A novel efficient Transformer Tuning (eTT) method that facilitates finetuning ViTs in the FSL tasks and proposes a novel prototypical regularization, which maximizes the similarity between the projected distribution of prefix and initial prototypes, regularizing the update procedure.
- ME-D2N: Multi-Expert Domain Decompositional Network for Cross-Domain Few-Shot Learning27
To solve the data imbalance problem between the source data with sufficient examples and the auxiliary target data with limited examples, a novel Multi-Expert Domain Decompositional Network (ME-D2N) is built under the umbrella of multi-expert learning.
- 27
- 27
- H4D: Human 4D Modeling by Learning Neural Compositional Representation26
Extensive experiments demonstrate the method presented is not only efficacy in recovering dynamic human with accurate motion and detailed geometry, but also amenable to various 4D human related tasks, including motion retargeting, motion completion and future prediction.
- 26
- EDEN: Enhanced Diffusion for High-quality Large-motion Video Frame Interpolation25
EDEN, an Enhanced Diffusion for high-quality large-motion vidEo frame iNterpolation, first utilizes a transformer-based tokenizer to produce refined latent representations of the intermediate frames for diffusion models, and enhances the diffusion transformer with temporal attention across the process and incorporates a start-end frame difference embedding to guide the generation of dynamic motion.
- Progressive Deep Neural Networks Acceleration via Soft Filter Pruning25
The proposed Progressive Soft Filter Pruning method prunes the network progressively and enables the pruned filters to be updated when training the model after pruning, which has three advantages over previous works: larger model capacity, less dependence on the pre-trained model and better performance.
- Vocabulary-Informed Zero-Shot and Open-Set Learning24
A weighted maximum margin framework for semantic manifold-based recognition that incorporates distance constraints from (both supervised and unsupervised) vocabulary atoms that ensures that labeled samples are projected closer to their correct prototypes, in the embedding space, than to others.
- Learning Large Euclidean Margin for Sketch-based Image Retrieval.24
A novel loss function, named Euclidean Margin Softmax (EMS), is proposed that not only minimizes intra-class distances but also maximizes inter- class distances simultaneously simultaneously, enabling us to learn a feature space with high discriminability, leading to highly accurate retrieval.
- UniDB: A Unified Diffusion Bridge Framework via Stochastic Optimal Control23
UniDB, a unified framework for diffusion bridges based on Stochastic Optimal Control based on Stochastic Optimal Control, achieves an optimal balance between control costs and terminal penalties, substantially improving detail preservation and output quality.
- DST: Dynamic Substitute Training for Data-free Black-box Attack23
A dynamic substitute structure learning strategy is proposed to adaptively generate optimal substitute model structure via a dy-namic gate according to different target models and tasks, which can achieve better performance compared with the state-of-the-art competitors on several datasets.
- Stacked Self-Attention Networks for Visual Question Answering23
A VQA model that utilizes the stacked self-attention for visual understanding, and the BERT-based question embedding model that enables the model to not only focus on a simple object but also the relations between objects is proposed.
- Learning from Synthetic Data Using a Stacked Multichannel Autoencoder23
This paper investigates and formalizes a general framework -- Stacked Multichannel Autoencoder (SMCAE) that enables bridging the synthetic gap and learning from synthetic data more efficiently and shows that this SMCAE can not only transform and use synthetic data on the challenging face-sketch recognition task, but that it can also help simulate real images, which can be used for training classifiers for recognition.
- LongVie: Multimodal-Guided Controllable Ultra-Long Video Generation22
LongVie, an end-to-end autoregressive framework for controllable long video generation, introduces two core designs to ensure temporal consistency: a unified noise initialization strategy that maintains consistent generation across clips, and global control signal normalization that enforces alignment in the control space throughout the entire video.
- 22
- 22
- Reinforcing Generated Images via Meta-Learning for One-Shot Fine-Grained Visual Recognition22
A meta-learning framework to combine generated images with original images, so that the resulting “hybrid” training images improve one-shot learning and show more diversity compared to the original and GAN-generated images.
- 22
- Self-supervised Amodal Video Object Segmentation21
A novel self-supervised learning paradigm that efficiently utilizes the visible object parts as the supervision to guide the training on videos, which achieves the state-of-the-art performance on the synthetic amodal segmentation benchmark FISHBOWL and the real world benchmark KINS-Video-Car.
- ManiTrans: Entity-Level Text-Guided Image Manipulation via Token-wise Semantic Alignment and Generation21
This work proposes a new transformer-based framework based on the two-stage image synthesis method, namely ManiTrans, which can not only edit the appearance of entities but also generate new entities corresponding to the text guidance.
- PourIt!: Weakly-supervised Liquid Perception from a Single Image for Visual Closed-Loop Robotic Pouring20
A simple yet effective framework PourIt!, to serve as a tool for robotic pouring tasks by designing a simple data collection pipeline that only needs image-level labels to reduce the reliance on tedious pixel-wise annotations.
- fMRI-PTE: A Large-scale fMRI Pretrained Transformer Encoder for Multi-Subject Brain Activity Decoding20
This work proposes fMRI-PTE, an innovative auto-encoder approach for fMRI pre- training, with a focus on addressing the challenges of varying fMRI data dimensions due to individual brain differences, and introduces a novel learning strategy tailored for pre-training 2D fMRI images, enhancing the quality of reconstruction.
- 20
- 20
- 19
- Frame-Transformer Emotion Classification Network19
A new architecture--Frame-Transformer Emotion Classification Network (FT-EC-net) is proposed to solve three highly correlated emotion analysis tasks: emotion recognition, emotion attribution and emotion-oriented summarization.
- 18
- Multi-view Shape Generation for a 3D Human-like Body18
This article proposes a multi-view shape generation network for a 3D human-like body reconstruction via a single RGB image using a coarse-to-fine learning model that gradually deforms a template body toward the ground truth body.
- Specialized re-ranking: A novel retrieval-verification framework for cloth changing person re-identification18
A novel retrieval-verification framework that will compare the input image and the candidate images by contrasting those local details and give a similarity score and its capability in improving the state-of-the-art methods remarkably on both synthetic and realistic datasets is proposed.
- 17
- 17
- Faster OreFSDet: A lightweight and effective few-shot object detector for ore images17
This work proposes a lightweight and effective few-shot detector to achieve competitive performance with general object detection with only a few samples for ore images, and achieves the smallest model size as well as being competitive at 50 FPS detection speed compared with General object detectors.
- Domain-Aware SE Network for Sketch-based Image Retrieval with Multiplicative Euclidean Margin Softmax17
A Domain-Aware Squeeze-and-Excitation (DASE) network is presented, which seamlessly incorporates the prior knowledge of sample sketch or photo into SE module and make the SE module capable of emphasizing appropriate channels according to domain signal.
- Comp-GAN17
A novel end-to-end Compositional Generative Adversarial Network that is able to synthesize new face images with specified poses and desired facial expressions and which can dynamically change the expression and pose of faces according to the input images while keeping the identity information.
- Knockoffs-SPR: Clean Sample Selection in Learning With Noisy Labels16
A novel theoretically guaranteed clean sample selection framework for learning with noisy labels by proposing a Scalable Penalized Regression with Knockoff filters (Knockoffs-SPR), which is provable to control the False-Selection-Rate (FSR) in the selected clean data.
- 16
- HOP: Heterogeneous Topology-based Multimodal Entanglement for Co-Speech Gesture Generation15
This work proposes a novel multimodal learning method named HOP for co-speech gesture generation that captures the heterogeneous entanglement between gesture motion, audio rhythm, and text semantics, enabling the generation of coordinated gestures.
- Disentangle Identity, Cooperate Emotion: Correlation-Aware Emotional Talking Portrait Generation15
Qualitative results and user studies further confirm the DICE-Talk method's ability to generate identity-preserving portraits with rich, correlated emotional expressions that naturally adapt to unseen identities, outperforming state-of-the-art approaches in emotion accuracy while maintaining competitive lip-sync performance.
- V$^{2}$-SAM: Marrying SAM2 with Multi-Prompt Experts for Cross-View Object Correspondence15
V2-SAM is presented, a unified cross-view object correspondence framework that adapts SAM2 from single-view segmentation to cross-view correspondence through two complementary prompt generators and introduces a Post-hoc Cyclic Consistency Selector (PCCS) that adaptively selects the most reliable expert based on cyclic consistency.
- Object-Centric Multiple Object Tracking15
This paper proposes a video object-centric model for MOT that consists of an index-merge module that adapts the object-centric slots into detection outputs and an object memory module that builds complete object prototypes to handle occlusions.
- Unsupervised Open-Vocabulary Object Localization in Videos15
A method that first localizes objects in videos via a slot attention approach and then assigns text to the obtained slots is proposed, which is effectively the first unsupervised approach that yields good results on regular video benchmarks.
- Exploring Structural Sparsity of Deep Networks Via Inverse Scale Spaces15
A new approach based on differential inclusions of inverse scale spaces that can generate a family of models from simple to complex ones along the dynamics via coupling a pair of parameters, such that over-parameterized deep models and their structural sparsity can be explored simultaneously.
- Feature Deformation Meta-Networks in Image Captioning of Novel Objects15
The feature deformation meta-networks (FDM-net), which is trained on source data, and learn to adapt to the novel object features detected by the auxiliary detection model, and the results show the effectiveness of the FDM-nets.
- ContextualStory: Consistent Visual Storytelling with Spatially-Enhanced and Storyline Context14
This work proposes ContextualStory, a novel framework designed to generate coherent story frames and extend frames for visual storytelling, which significantly outperforms existing SOTA methods in both story visualization and continuation.
- Content and Salient Semantics Collaboration for Cloth-Changing Person Re-Identification14
This paper proposes a unified Semantics Mining and Refinement module to extract robust identity-related content and salient semantics, mitigating interference from clothing appearances effectively, and proposes the Content and Salient Semantics Collaboration framework to collaborate and leverage various semantics, facilitating cross-parallel semantic interaction and refinement.
- 14
- Polaris: Open-ended Interactive Robotic Manipulation via Syn2Real Visual Grounding and Large Language Models14
This paper introduces an interactive robotic manipulation framework called Polaris, which integrates perception and interaction by utilizing GPT-4 alongside grounded vision models, and proposes a novel Synthetic-to-Real (Syn2Real) pose estimation pipeline.
- 14
- Concept Replacer: Replacing Sensitive Concepts in Diffusion Models via Precision Localization13
Experimental results demonstrate that the proposed novel approach for targeted concept replacement in diffusion models achieves superior precision in localizing target concepts and performs coherent concept replacement with minimal impact on non-target areas, outperforming existing approaches.
- Image-Text-Image Knowledge Transfer for Lifelong Person Re-Identification With Hybrid Clothing States13
This paper proposes a novel framework, dubbed Teata, to effectively align, transfer, and accumulate knowledge in an “image-text-image” closed loop, and introduces a Knowledge Adaptation and Projection strategy, which tunes text knowledge via a slow-paced learner to adapt to different tasks without catastrophic forgetting.
- Rethinking Person Re-identification from a Projection-on-Prototypes Perspective13
This paper proposes a new baseline ProNet, which innovatively reserves the function of the classifier at the inference stage, and describes the identity of input images as similarities to all prototypes, which are then utilized as more discriminative features to perform person Re-ID.
- Whose hand is this? Person Identification from Egocentric Hand Gestures13
This paper systematically study the possibility of Egocentric Hand Identification (EHI) with unconstrained egocentric hand gestures, and shows that EHI can improve generalization of other tasks, such as gesture recognition, by training adversarially to encourage these models to ignore differences between users.
- 13
- 13
- Multi-view Metric Learning for Multi-view Video Summarization13
A multiview metric learning framework is presented that combines the advantages of maximum margin clustering with the disagreement minimization criterion and has the ability to find a metric that best separates the input data, and meanwhile to force the learned metric to maintain underlying intrinsic structure of data points.
- Sequential Multi-Object Grasping with One Dexterous Hand12
This paper proposes SeqMultiGrasp, a system for sequentially grasping objects with a four-fingered Allegro Hand, which focuses on sequentially grasping two objects, ensuring that the hand fully encloses one object before lifting it and then grasps the second object without dropping the first.
- Diffusion-Based Imaginative Coordination for Bimanual Manipulation12
A multi-frame latent prediction strategy that encodes future states in a compressed latent space, preserving task-relevant features and a unidirectional attention mechanism where video prediction is conditioned on the action, while action prediction remains independent of video prediction is introduced.
- Spatial-Temporal Aware Visuomotor Diffusion Policy Learning12
This work proposes 4D Diffusion Policy (DP4), a novel visual imitation learning method that incorporates spatiotemporal awareness into diffusion-based policies and leverages a dynamic Gaussian world model to guide the learning of 3D spatial and 4D spatiotemporal perceptions from interactive environments.
- 12
- 12
- Pushing the Limits of 3D Shape Generation at Scale12
This work developed a remarkable model with an astounding 3.6 billion trainable parameters, establishing it as the largest 3D shape generation model to date, named Argus-3D, and introduces a discrete codebook for efficient quantization of these representations.
- WALL-E: Embodied Robotic WAiter Load Lifting with Large Language Model12
The potential of integrating the most recent Large Language Models (LLMs) and existing visual grounding and robotic grasping system to enhance the effectiveness of the human-robot interaction is investigated.
- 12
- 6′-Hydroxy Justicidin B Triggers a Critical Imbalance in Ca2+ Homeostasis and Mitochondrion-Dependent Cell Death in Human Leukemia K562 Cells12
The results indicate that HJB inhibited the proliferation of K562 cells and induced apoptosis by affecting the function of mitochondria and calcium homeostasis to activate the p53 signaling pathway, presenting HJB as a potential novel alternative to standard human leukemia therapies.
- MinD-3D++: Advancing fMRI-Based 3D Reconstruction With High-Quality Textured Mesh Generation and a Comprehensive Dataset11
The fMRI-3D dataset is presented, a novel framework for decoding textured 3D visual information from fMRI signals is proposed, and the framework evaluates the feasibility of not only reconstructing 3D objects from the human mind but also generating, for the first time, 3D textured meshes with detailed textures from fMRI data.
- Pushing Auto-regressive Models for 3D Shape Generation at Capacity and Scalability11
This paper extends auto-regressive models to 3D domains, and introduces discrete representation learning based on a latent vector instead of volumetric grids, which not only reduces computational costs but also preserves essential geometric details by learning the joint distributions in a more tractable order.
- Neural Symbolic Representation Learning for Image Captioning11
The first attempt to investigate the ability of neural symbolic representation of the image for the image captioning task is made and it is demonstrated to better generalize to other domains with significant performance improvements compared with existing methods on the cross domain image captioned task.
- Dynamic Depth Fusion and Transformation for Monocular 3D Object Detection11
This work designs an Adaptive Depth-guided Instance Normalization layer to leverage depth features to guide RGB features for high quality estimation of 3D properties and introduces a Dynamic Depth Transformation module to better recover accurate depth according to semantic context learning.
- 11
- 11
- 11
- Brain-machine interactive neuromodulation research tool with edge AI computing10
This study developed a brain-machine interactive neuromodulation research tool (BMINT), which has capabilities of neurophysiological signals sensing, computing with mainstream machine learning algorithms and delivering electrical stimulation pulse by pulse in real-time.
- Rethinking Amodal Video Segmentation from Learning Supervised Signals with Object-centric Representation10
This paper proposes an Efficient object-centric Representation amodal Segmentation (EoRaS), and designs a translation module to project image features into the Bird’s-Eye View (BEV), which introduces 3D information to improve current feature quality.
- 10
- 9
- 9
- 9
- Clustering by the Probability Distributions From Extreme Value Theory9
A novel clustering algorithm, which models the distributions of distances to centroids over a threshold by the generalized Pareto distribution (GPD) in extreme value theory, and introduces a naive baseline, dubbed as generalized extreme value (GEV) <inline-formula><tex-math notation="LaTeX">$k$</tex- Math></inline- formula>-means.
- 9
- 9
- A Neural Representation Framework with LLM-Driven Spatial Reasoning for Open-Vocabulary 3D Visual Grounding8
This work proposes SpatialReasoner, a novel neural representation-based framework with large language model (LLM)-driven spatial reasoning that constructs a visual properties-enhanced hierarchical feature field for open-vocabulary 3D visual grounding.
- CAP-Net: A Unified Network for 6D Pose and Size Estimation of Categorical Articulated Parts from a Single RGB-D Image8
This paper proposes a single-stage Network, CAP-Net, for estimating the 6D poses and sizes of Categorical Articulated Parts, and introduces the RGBD-Art dataset, the largest RGB-D articulated dataset to date, featuring photorealistic RGB images and depth noise simulated from real sensors.
- 8
- ImpDet: Exploring Implicit Fields for 3D Object Detection8
This work introduces a new perspective that views bounding box regression as an implicit function, which leverages implicit field learning for 3D object detection and presents a simple yet efficient virtual sampling strategy to solve the problem of sparsity on the object surface.
- A Simple Test-Time Method for Out-of-Distribution Detection8
A simple Test-time Linear Training (ETLT) method for OOD detection, which achieves promising performance and is more practical in real-world applications, and an online variant of the proposed method, which achieves promising performance and is more practical in real-world applications.
- 8
- 8
- Learning to Separate Domains in Generalized Zero-Shot and Open Set Learning: a probabilistic perspective8
A probabilistic way of directly estimating and fine-tuning the decision boundary between seen and novel/unseen classes is proposed and the uncertain domain is newly introduced in this framework to adopt those instances whose domain cannot be predicted confidently.
- BigVid at MediaEval 2016: Predicting Interestingness in Images and Videos.8
Results show that the visual features and high-level attributes are complementary to each other, which is interesting and potential useful in the MediaEval 2016 Predicting Media Interestingness Task.
- 8
- SwiftVideo: A Unified Framework for Few-Step Video Generation through Trajectory-Distribution Alignment7
This work proposes SwiftVideo, a unified and stable distillation framework that combines the advantages of trajectory-preserving and distribution-matching strategies, and introduces continuous-time consistency distillation to ensure precise preservation of ODE trajectories.
- 3D StreetUnveiler with Semantic-aware 2DGS -- a simple baseline7
A novel time-reversal framework to inpaint frames in reverse order and use later frames as references for earlier frames to fully utilize the long-trajectory observations of the empty street.
- LAC-Net: Linear-Fusion Attention-Guided Convolutional Network for Accurate Robotic Grasping Under the Occlusion7
This paper introduces a new framework that explores amodal segmentation for robotic grasping in cluttered scenes, thus greatly enhancing robotic grasping abilities and proposing a Linear-fusion Attention-guided Convolutional Network (LAC-Net).
- Repositioning the Subject within Image7
This research reveals that the fundamental sub-tasks of subject repositioning can be effectively reformulated as a unified, prompt-guided inpainting task, and forms the basis of the SEgment-gEnerate-and-bLEnd (SEELE) framework.
- Doubly Robust Proximal Causal Learning for Continuous Treatments7
This work proposes a kernel-based DR estimator that can well handle continuous treatments and shows that its oracle form is a consistent approximation of the influence function, and proposes a new approach to efficiently solve the nuisance functions.
- Split-PU: Hardness-aware Training Strategy for Positive-Unlabeled Learning7
This paper focuses on improving the commonly-used nnPU with a novel training pipeline and highlights the intrinsic difference of hardness of samples in the dataset and the proper learning strategies for easy and hard data.
- 7
- 7
- VTBench: Comprehensive Benchmark Suite Towards Real-World Virtual Try-on Models6
VTBench is a hierarchical benchmark suite that systematically decomposes virtual image try-on into hierarchical, disentangled dimensions, each equipped with tailored test sets and evaluation criteria, to foster the field of virtual try-on towards challenging real-world scenario.
- 6
- 6
- 6
- SparseGrasp: Robotic Grasping via 3D Semantic Gaussian Splatting from Sparse Multi-View RGB Images6
Experimental results show that SparseGrasp significantly outperforms state-of-the-art methods in terms of both speed and adaptability, providing a robust solution for multi-turn grasping in changeable environment.
- 6
- 6
- Adaptive Pruning of Pretrained Transformer via Differential Inclusions5
To achieve effective pruning, the transformers are segmented into paired modules, including query-key pairs, value-projection pairs, and sequential linear layers, and apply low-rank compression to these pairs, maintaining the output structure while enabling structural compression within the inner states.
- 5
- Making Your Dreams A Reality: Decoding the Dreams into a Coherent Video Story from fMRI Signals5
A novel approach to visually decoding dreams using fMRI signals and weaving dream visuals into narratives using language models is introduced, to push the boundaries of dream research and gain deeper insights into visual experiences during sleep.
- You Only Estimate Once: Unified, One-stage, Real-Time Category-Level Articulated Object 6D Pose Estimation for Robotic Grasping5
YOEO is proposed, a single-stage method that simultaneously outputs instance segmentation and NPCS representations in an end-to-end manner and deploys a synthetically-trained model in a real-world setting, providing real-time visual feedback at 200Hz, enabling a physical Kinova robot to interact with unseen articulated objects.
- Cross-View Multi-Modal Segmentation @ Ego-Exo4D Challenges 20255
A cross-view multi-modal object segmentation approach that enhances object localization by leveraging both visual masks and textual descriptions as segmentation conditions and introduces a cross-view object alignment module that enforces object-level consistency across perspectives, thereby improving the model's robustness to viewpoint changes.
- TC‐GEN: Data‐Driven Tropical Cyclone Downscaling Using Machine Learning‐Based High‐Resolution Weather Model5
This approach consists of multiple components: a data‐driven model for generating synthetic TC seeds, a blending method that seamlessly integrates storm seeds into the surrounding while maintaining the seed structure, and a model based on a recurrent neural network to correct for biases in storm intensity.
- Local Consensus Enhanced Siamese Network with Reciprocal Loss for Two-view Correspondence Learning5
This work proposes a Local Feature Consensus (LFC) plugin block to augment the features of existing models and extends existing models to a Siamese network with a reciprocal loss that exploits the supervision of mutual projection, which considerably promotes the matching performance without introducing additional model parameters.
- Joint fMRI Decoding and Encoding with Latent Embedding Alignment5
This work proposes an end-to-end framework, SemanSig, which directly encodes fMRI signals and extracts semantic information, and uses the ImageNet class prototype space as the internal representation space of fMRI signals, thereby reducing signal redundancy and learning difficulty.
- Online Dense Point Tracking with Streaming Memory4
This work presents a lightweight and fast model with Streaming memory for dense POint Tracking and online video processing that enables SPOT to perform dense point tracking with state-of-the-art accuracy on the CVO benchmark, as well as comparable or superior performance to offline models on sparse tracking benchmarks such as TAP-Vid and RoboTAP.
- ReVSeg: Incentivizing the Reasoning Chain for Video Segmentation with Reinforcement Learning4
This work introduces ReVSeg, which executes reasoning as sequential decisions in the native interface of pretrained vision language models (VLMs), and employs reinforcement learning to optimize the multi-step reasoning chain, enabling the model to self-refine its decision quality from outcome-driven signals.
- Improving Neural Surface Reconstruction with Feature Priors from Multi-View Image4
This study comprehensively explore multi-view feature priors from seven pretext visual tasks, revealing that feature priors from image matching and multi-view stereo outperform other pretext tasks and discovering that extending patch-wise photometric consistency to the feature level surpasses the performance of pixel-wise approaches.
- Entity-Level Text-Guided Image Manipulation4
This work proposes an elegant framework, dubbed as SeMani, forming the Semantic Manipulation of real-world images that can not only edit the appearance of entities but also generate new entities corresponding to the text guidance.
- Open-DDVM: A Reproduction and Extension of Diffusion Model for Optical Flow Estimation4
This technical report presents the first open-sourceDDVM by reproducing it and achieves comparable performance to the closed-source DDVM by training on 40k public data with 4 GPUs.
- 4
- Regularising Knowledge Transfer by Meta Functional Learning4
Experiments show that meta functional learning for regularisation knowledge transfer can benefit improving FSL classifiers and adopt an Iterative Update strategy on MFL (MFL-IU), which improves knowledge transfer regularisation from MFL by progressively learning the functional regularisation in knowledge transfer.
- Parsimonious Deep Learning: A Differential Inclusion Approach with Global Convergence.4
A parsimonious learning approach based on differential inclusions of inverse scale spaces, that generates a family of models from simple to complex ones with a better efficiency and interpretability than stochastic gradient descent in exploring the model space is proposed.
- Stacked multichannel autoencoder – an efficient way of learning from synthetic data4
This paper investigates and formalizes a general framework – Stacked Multichannel Autoencoder (SMCAE) that enables bridging the synthetic gap and learning from synthetic data more efficiently and shows that this SMCAE can not only transform and use synthetic data on a challenging face-sketch recognition task, but that it can also help simulate real images which can be used for training classifiers for recognition.
- 4
- 3
- DecoFuse: Decomposing and Fusing the "What", "Where", and "How" for Brain-Inspired fMRI-to-Video Decoding3
DecoFuse is proposed, a novel brain-inspired framework for decoding videos from fMRI signals that first decomposes the video into three components - semantic, spatial, and motion - then decodes each component separately before fusing them to reconstruct the video.
- CustAny: Customizing Anything from A Single Example3
This paper proposes a novel pipeline to construct a large dataset of general objects and build the Multi-Category ID-Consistent (MC-IDC) dataset, and introduces Customizing Anything (CustAny), a zero-shot framework that maintains ID fidelity and supports flexible text editing for general objects.
- EFCNet: Every Feature Counts for Small Medical Object Segmentation3
A novel model named EFCNet is proposed for small object segmentation in medical images that addresses information loss during encoding and decoding procedures and significantly outperforms previous segmentation methods designed for both medical and normal images.
- 3
- 3
- Learning Versatile 3D Shape Generation with Improved AR Models3
The Improved Auto-regressive Model (ImAM) is proposed for 3D shape generation, which applies discrete representation learning based on a latent vector instead of volumetric grids and can synthesize diverse and faithful shapes of multiple categories, achieving state-of-the-art performance.
- HandO: a hybrid 3D hand–object reconstruction model for unknown objects3
A novel hybrid model—hand–object Model (HandO) enabling the hand–object 3D reconstruction with smooth surfaces and accurate positions is presented, for the first time making the hybrid 3D representation for this task by bringing meshes, SDFs, and parametric models together.
- Local Slot Attention for Vision and Language Navigation3
This work proposes a slot-attention based module to incorporate information from segmentation of the same object in the VLN task and a local attention mask mechanism to limit the visual attention span.
- Second Order Enhanced Multi-glimpse Attention in Visual Question Answering3
A novel Second Order enhanced Multi-glimpse Attention model (SOMA) where each glimpse denotes an attention map and an explicit second order feature is constructed to model the interaction on both the intra-modality and cross-modality of features.
- Can Action be Imitated? Learn to Reconstruct and Transfer Human Dynamics from Videos3
This paper introduces a novel task, dubbedmeshbased action imitation, to enable an arbitrary target human mesh to perform the same action shown on the video demonstration, using a novel Mesh-based Video Action Imitation (M-VAI) method.
- Data-efficient Alignment of Multimodal Sequences by Aligning Gradient Updates and Internal Feature Distributions3
This pa-per attempts to enhance data efficiency of the end-to-end alignment network NeuMATCH by employing layer-wise adaptive rate scaling (LARS) and sequence-wise batch normalization (SBN) to align the internal feature distributions from different modalities.
- A New Screening Method for COVID-19 based on Ocular Feature Recognition by Machine Learning Tools3
A new screening method of analyzing the eye-region images, captured by common CCD and CMOS cameras, could reliably make a rapid risk screening of COVID-19 with very high accuracy and a system implementing such an algorithm should assist the triage management or the clinical diagnosis.
- 3
- 2
- RAG-6DPose: Retrieval-Augmented 6D Pose Estimation via Leveraging CAD as Knowledge Base2
RAG-6DPose is presented, a retrieval-augmented approach that leverages 3D CAD models as a knowledge base by integrating both visual and geometric cues, demonstrating the effectiveness and robustness of the approach, particularly in handling occlusions and novel viewpoints.
- Learning a Mixture of Conditional Gating Blocks for Visual Question Answering2
A novel model mixture of conditional gating blocks (McG) is proposed for VQA, which keeps the best of the Transformer, convolutional neural network (CNN), and dynamic networks.
- A Generalization Theory of Cross-Modality Distillation with Contrastive Learning2
This work forms a general framework of cross-modality contrastive distillation (CMCD), built upon contrastive learning that leverages both positive and negative correspondence, towards a better distillation of generalizable features.
- 2
- 2
- 2
- 2
- Wavelet Prior Attention Learning in Axial Inpainting Network2
A novel model -- Wavelet prior attention learning in Axial Inpainting Network (WAIN), whose generator contains the encoder, decoder, as well as two key components of Wavelet image Prior Attention (WPA) and stacked multi-layer Axial-Transformers (ATs).
- 2
- 2
- Attribute Learning for Image/Video Understanding2
A framework for learning latent attributes is present for automatic classification and annotation of unstructured group social activity in videos, which enables the tasks of attribute learning for understanding complex multimedia data with sparse and incomplete labels.
- 2
- 2
- A new formulation of Lipschitz constrained with functional gradient learning for GANs1
It is demonstrated that the neighborhood size of the latent vector can be reduced by increasing the norm of the discriminator gradient, resulting in enhanced diversity of synthetic samples.
- Ali-UI: Enhancing Complex Vision-Language Navigation with Alignment of Unified Map and Instruction Parsing1
Ali-UI is introduced, a novel vision-language navigation approach that enables agents to navigate from random starting points in unvisited scenes and handle complex multi-step instructions by incorporating continuously accumulating global grid maps and local semantic maps as scene memory by employing frontier-based exploration.
- 1
- 1
- 1
- 1
- 1
- When Preferences Diverge: Aligning Diffusion Models with Minority-Aware Adaptive DPO1
An Adaptive-DPO loss function is introduced which improves the DPO loss in two ways: enhancing the model's learning of majority labels while mitigating the negative impact of minority samples, paving the way for more effective training methodologies in image generation tasks.
- Hyper-Transformer for Amodal Completion1
This work introduces a novel framework named the Hyper-Transformer Amodal Network (H-TAN), which utilizes a hyper transformer equipped with a dynamic convolution head to directly learn shape priors and accurately predict amodal masks.
- Intelligent Director: An Automatic Framework for Dynamic Visual Composition using ChatGPT1
An Intelligent Director framework is proposed, utilizing LENS to generate descriptions for images and video frames and combining ChatGPT to generate coherent captions while recommending appropriate music names, which aims to automatically integrate various media elements based on user requirements and create storytelling videos.
- 1
- A deep learning-based prognostic framework for aeroengine exhaust gas temperature margin1
An attention-based deep learning framework was proposed for the aeroengine performance prediction model, and the proposed attention mechanism is adopted to learn the influence of features and time steps, assigning different weights according to their importance.
- Class-Incremental Generalized Zero-Shot Learning1
Experiments on several benchmarks show that DPL has the capacity of remembering the knowledge learned from previous source instances and be able to migrate all the knowledge to the target domain as the expanding of the source domain.
- 1
- A Framework of Meta Functional Learning for Regularising Knowledge Transfer1
This work proposes a novel framework of Meta Functional Learning by meta-learning a generalisable functional model from data-rich tasks whilst simultaneously regularising knowledge transfer to data-scarce tasks and shows that meta functional learning for knowledge transfer regularisation can improve FSL classifiers.
- 1
- 1
- Incrementally Zero-Shot Detection by an Extreme Value Analyzer1
This paper proposes a novel extreme value analyzer to detect objects from old seen, new seen, and unseen classes, simultaneously, and proposes two innovative losses, i.e., background-foreground mean squared error loss alleviating the extreme imbalance of the background and foreground of images, and projection distance loss aligning the visual space and semantic spaces of old seen classes.
- 1
- 1
- Extreme vocabulary learning1
A novel visual-semantic embedding framework — extreme vocabulary learning (EVoL) is proposed; the EVoL embeds the visual features into semantic space in a probabilistic way and adopts the vast open vocabulary in the semantic space to help further constraint the margin and coverage of training classes.
- 1
- 1
- Learning to score and summarize figure skating sport videos.1
A reinforcement learning based video summarization algorithm is proposed here; and the experiments show better performance than the other baseline video summarizing algorithms.
- 1
- Left-Right Skip-DenseNets for Coarse-to-Fine Object Categorization.1
The left-right skip-densenets for coarse-to-fine object categorization can enable both coarse and fine-grained classification in a single framework and the layer-skipping mechanism which learns a gating network to predict whether skip some layers in the testing stage is proposed.
- 1
- –
- Cross-subject fMRI-to-Image with Visual-cortex 2D Representation and Pre-Training–
X-MinD is presented, a novel zero-shot cross-subject fMRI-to image approach designed to reconstruct images from brain activity across unseen subjects that achieves effective zero-shot decoding of cross-subject fMRI activity into realistic and semantically consistent images, bypassing the need for model fine tuning.
- –
- –
- –
- –
- UniPruning: Unifying Local Metric and Global Feedback for Scalable Sparse LLMs–
UniPruning is presented, a unified post-training pruning framework that combines the speed of local saliency metrics with the stability of global coordination, enabled by a mirror descent based optimization, all without updating model weights.
- SafeCtrl: Region-Based Safety Control for Text-to-Image Diffusion via Detect-Then-Suppress–
SafeCtrl is introduced, a lightweight, non-intrusive plugin that first precisely localizes unsafe content and suppresses the harmful semantics, allowing the generative process to naturally and coherently resolve into a safe, context-aware alternative.
- –
- –
- –
- COSMO-RL: Towards Trustworthy LMRMs via Joint Safety and Stability–
COSMO-RL is presented, a mixed reinforcement learning framework that trains reasoning oriented LMRMs under multimodal, multitask, and multiobjective signals, and the resulting model, COSMO-R1, improves safety while maintaining-and often improving multimodal reasoning and instruction following.
- EFCNet enhances the efficiency of segmenting clinically significant small medical objects–
EFCNet is introduced, which integrates the Cross-Stage Axial Attention (CSAA) module for enhanced feature fusion and the Multi-Precision Supervision (MPS) module for improved hierarchical guidance and outperformed state-of-the-art models where conventional models underperform.
- Financial Performance of Innovation Investment in SpecializedSMEs Under A Digital Business Environment: A ComparativeAnalysis of China and Thailand (2019-2024)–
This is the first comprehensive comparative study examining the financial performance of innovation investment in China and Thailand's specialized SMEs for the critical digital transformation period of 2019-2024, providing theoretical and practical implications to emerging market SMEs.
- Robust Network Learning via Inverse Scale Variational Sparsification–
This work introduces an inverse scale variational sparsification framework within a time-continuous inverse scale space formulation that progressively learns finer-scale features by discerning variational differences between pixels, ultimately preserving only large-scale features in the smoothed image.
- –
- –
- –
- –
- Rethinking the Multi-view Stereo from the Perspective of Rendering-based Augmentation–
This work applies one of the state-of-the-art learning-based MVS methods, --MVSFormer, to overcome intractable scenarios such as textureless and reflections regions suffered by traditional PatchMatch methods, and finds that the MVS method could produce much better predictions through rendered images due to the coincident illumination, which is significant for the M VS community.
- –
- –
- The Report on China-Spain Joint Clinical Testing for Rapid COVID-19 Risk Screening by Eye-region Manifestations–
A new fast screening method of analyzing the eye-region images, captured by common CCD and CMOS cameras, which could reliably make a rapid risk screening of COVID-19 with the sustainable stable high performance in different countries and races.
- Report on Three Round COVID-19 Risk Blind Tests by Screening Eye-region Manifestations–
A new fast screening method of analyzing the eye-region images, captured by common CCD and CMOS cameras, which could reliably make a rapid risk screening of COVID-19 with the sustainable stable high performance in different countries and races.
- A COVID-19 screening method based on eyes photographs and artificial intelligence–
Preliminary results show that this DLS performed well in identifying probable asymptomatic COVID-19 cases through the analysis of participants' eyes pictures, suggesting that this method could be an innocuous, accessible, low cost and quick CO VID-19 screening method.
- –
- Learning decomposed subspaces for supervised bidirectional image generation–
A framework to learn a feature space, which breaks into three parts, namely, two personalised semantic subspaces, which encode patterns that are unique for each domain, and a shared semantic subspace that captures the common patterns.
- –
- –
- –
- –
- –
- –
- –
- –
- VOCABULARY-INFORMED VISUAL FEATURE AUGMENTATION FOR ONE-SHOT LEARNING–
An extensive study shows that data augmented from semantic space are complementary with those from the image space, and thus boost the classification accuracy dramatically and effectively improves the performance of one-shot classification.
- –
- VECTORIZATION METHODS IN RECOMMENDER SYSTEM–
The similarity based on cooccurrence information suffers from cold start, and a content based similarity model based on doc2vec which is another technology in NLP is proposed and reach state-of-art in recommender system.
- –
- –
- –
- Construction of the New Comprehensive Curriculum Design to Speed up the Culture of the Creative Thinking–
The article combines the comprehensive curriculum design and the culture of creative thinking to show that using information technology, constructing emulate project environment, students can be able to use factual scene media, principle media and network media so that the promotion of creativethinking can be achieved.
- The Development of IC Card Terminal System–
The general design principle is that IC card application system and the particulars should be noticed when the card is applied in the campus management system.
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 2026-10-12. 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.