Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks137 from public data
- MovieChat: From Dense Token to Sparse Memory for Long Video Understanding685
The MovieChat achieves state-of-the-art performance in long video understanding, along with the released MovieChat-1K benchmark with 1K long video and 14K manual annotations for validation of the effectiveness of the effectiveness of the method.
- StableVideo: Text-driven Consistency-aware Diffusion Video Editing243
This paper develops a novel inter-frame propagation mechanism for diffusion video editing, which leverages the concept of layered representations to propagate the appearance information from one frame to the next.
- Exploit the Connectivity197
This paper proposes an innovative and effective tracking method called TrackletNet Tracker (TNT) that combines temporal and appearance information together as a unified framework and achieves promising results on MOT16 and MOT17 benchmark datasets compared with other state-of-the-art methods.
- 156
- AI Assisted Fashion Design: A Review113
The primary results highlight the transformative potential of enhanced personalization and seamless multimodal interfaces, empowering designers and consumers to co-create unique and personalized designs within the fashion industry.
- Multi-Camera Tracking of Vehicles based on Deep Features Re-ID and Trajectory-Based Camera Link Models110
An MCT system, which combines single- camera tracking (SCT) and inter-camera tracking (ICT) which includes trajectory-based camera link model and deep feature reidentification and is evaluated on CVPR AI City Challenge 2019 City Flow dataset, achieving IDF1 70.59%, which outperforms competing methods.
- MovieChat+: Question-Aware Sparse Memory for Long Video Question Answering85
This work proposes MovieChat within a training-free memory consolidation mechanism to overcome challenges in long video understanding, and lifts pre-trained large multi-modal models for understanding long videos without additional trainable modules, employing a zero-shot approach.
- Deep learning-enabled 3D multimodal fusion of cone-beam CT and intraoral mesh scans for clinically applicable tooth-bone reconstruction85
Deep Dental Multimodal Fusion is presented, an automatic multimodal framework that reconstructs 3D tooth-bone structures using CBCT and IOS using a novel pixel representation learning architectures, prior knowledge-guided losses, and geometry-based 3D fusion techniques.
- Beware of the generic machine learning-based scoring functions in structure-based virtual screening81
Overall, generic MLSFs in VS could hardly achieve satisfactory results for any dataset, and they could even not outperform the baseline of classical SFs such as Glide SP.
- Bridging Cross-task Protocol Inconsistency for Distillation in Dense Object Detection75
A novel distillation method with cross-task consistent protocols, tailored for the dense object detection, and an IoU-based Localization Distillation Loss that is free from specific network structures and can be compared with existing localization distillation losses are proposed.
- Track without Appearance: Learn Box and Tracklet Embedding with Local and Global Motion Patterns for Vehicle Tracking74
A novel approach is proposed that tackles the association issue for long-term tracking with the exclusive fully-exploited motion information with the proposed reconstruct-to-embed strategy based on deep graph convolutional neural networks (GCN).
- Language Adaptive Weight Generation for Multi-Task Visual Grounding73
An active perception Visual Grounding framework based on Language Adaptive Weights, called VG-LAW, which serves as an expression-specific feature extractor through dynamic weights generated for various expressions and does not require additional modules for cross-modal interaction.
- Can machine learning consistently improve the scoring power of classical scoring functions? Insights into the role of machine learning in scoring functions73
A comparative assessment of 25 commonly used classical SFs was conducted, and the results indicated that the superiority of the ML-based SFs could be fully guaranteed when sufficient similar targets were contained in the training set.
- DiffFashion: Reference-Based Fashion Design With Structure-Aware Transfer by Diffusion Models68
This work presents a novel diffusion model-based unsupervised structure-aware transfer method, namely DiffFashion, which outperforms state-of-the-art baseline models, generating more realistic images in the fashion design task.
- Rethinking of Radar’s Role: A Camera-Radar Dataset and Systematic Annotator via Coordinate Alignment68
A new dataset, named CRUW, with a systematic annotator and performance evaluation system to address the radar object detection (ROD) task, which aims to classify and localize the objects in 3D purely from radar’s radio frequency (RF) images.
- 67
- STSC-SNN: Spatio-Temporal Synaptic Connection with temporal convolution and attention for spiking neural networks61
This work takes inspiration from biological synapses and proposes a Spatio-Temporal Synaptic Connection SNN (STSC-SNN) model to enhance the spatio-temporal receptive fields of synaptic connections, thereby establishing temporal dependencies across layers.
- Split and Connect: A Universal Tracklet Booster for Multi-Object Tracking59
Combined with the proposed tracklet booster, existing trackers can achieve large improvements on the IDF1 score, which shows the effectiveness of the proposed TBooster.
- Anomaly Candidate Identification and Starting Time Estimation of Vehicles from Traffic Videos55
A novel two-stage framework for anomaly event detection in road traffic based on anomaly candidate identification and starting time estimation of vehicles is proposed.
- 54
- MedM2G: Unifying Medical Multi-Modal Generation via Cross-Guided Diffusion with Visual Invariant52
MedM2G is the first medical generative model that unifies medical generation tasks of text-to-image, image-to-text, and unified generation of medical modalities (CT, MRI, X-ray), and it performs 5 medical generation tasks across 10 datasets, consistently outperforming various state-of-the-art works.
- Sustainable Road Pothole Detection: A Crowdsourcing Based Multi-Sensors Fusion Approach52
This study proposes a method that utilizes the fusion of accelerometer data and in-vehicle video data, which is uploaded by the participating users, and provides a novel approach for large-scale pavement anomaly detection.
- DIVOTrack: A Novel Dataset and Baseline Method for Cross-View Multi-Object Tracking in DIVerse Open Scenes49
The DIVOTrack is introduced: a novel baseline cross-view tracking method with a unified joint detection and cross-view tracking framework named CrossMOT, which learns object detection, single-view association, and cross-view matching with an all-in-one embedding model.
- Deep Vision Multimodal Learning: Methodology, Benchmark, and Trend48
The types of architectures used in multimodal learning, including feature extraction, modality aggregation, and multi-modality loss functions, are reviewed, and it is indicated that pretraining paradigm, unified multitask framework, missing and noisy modality, and multimodale task diversity could be the future trends and challenges in the deep vision multimodAL learning field.
- 47
- 46
- A Survey of Deep Learning in Sports Applications: Perception, Comprehension, and Decision40
A comprehensive survey of deep learning in sports performance, focusing on three main aspects: algorithms, datasets and virtual environments, and challenges, provides valuable reference material for researchers interested in deep learning in sports applications.
- Uncertainty-Based Active Learning via Sparse Modeling for Image Classification39
This paper uses sparse linear combination to represent the uncertainty of unlabeled pool data with Gaussian kernels, in which the diversity and density are well incorporated, and proposes a novel method that combines uncertainty, diversity, and density via sparse modeling in the sample selection.
- Citygen: Infinite and Controllable City Layout Generation37
This paper presents CityGen, an end-to-end framework for infinite, diverse, and controllable city layout generation, and introduces an infinite expansion module to extend local layouts to city-scale layouts and a multi-scale refinement module to upsample and refine them.
- LLaVA-Ultra: Large Chinese Language and Vision Assistant for Ultrasound37
This paper proposes a fine-grained adaptive VLM architecture for Chinese medical visual conversations through parameter-efficient tuning, and devise a fusion module with fine-grained vision encoders to achieve enhancement for subtle medical visual semantics.
- CityCraft: A Real Crafter for 3D City Generation37
The CityCraft approach integrates three key stages: initially, a diffusion transformer model is deployed to generate diverse and controllable 2D city layouts, and a Large Language Model (LLM) is utilized to strategically make land-use plans within these layouts based on user prompts and language guidelines.
- PoSynDA: Multi-Hypothesis Pose Synthesis Domain Adaptation for Robust 3D Human Pose Estimation33
The Multi-Hypothesis Pose Synthesis Domain Adaptation (PoSynDA) framework is presented, which simulates the 3D pose distribution in the target domain, filling the data diversity gap and paves the way for the practical application of 3D human pose estimation.
- Hierarchical Auto-Organizing System for Open-Ended Multi-Agent Navigation32
The HAS framework is designed to develop embodied organizations that push the boundaries of embodied AI, moving it towards a more human-like organizational structure and to assess organizational behavior, which includes searching and exploring in the Minecraft environment.
- Deep Learning Methods for Small Molecule Drug Discovery: A Survey30
This article presents a comprehensive review on molecule generation, molecular property prediction, retrosynthesis prediction, retrosynthesis prediction, and reaction prediction in deep learning, and discusses the relationships among different applications.
- 30
- 29
- BEVSpread: Spread Voxel Pooling for Bird's-Eye-View Representation in Vision-Based Roadside 3D Object Detection29
Instead of bringing the image features contained in a frustum point to a single BEV grid, BEVSpread considers each frustum point as a source and spreads the image features to the surrounding BEV grids with adaptive weights to achieve superior prop- agation performance.
- Global Adaptation meets Local Generalization: Unsupervised Domain Adaptation for 3D Human Pose Estimation29
PoseDA combines global adaptation and local generalization in PoseDA, a simple yet effective framework of unsupervised domain adaptation for 3D human pose estimation and proposes local pose augmentation (LPA) to enhance the diversity of 3D poses following an adversarial training scheme.
- Exploring Severe Occlusion: Multi-Person 3D Pose Estimation with Gated Convolution28
This paper proposes a temporal regression network with a gated convolution module to transform 2D joints to 3D and recover the missing occluded joints in the meantime and shows that the proposed method outperforms most state-of-the-art 2D-to-3D pose estimation methods, especially for the scenarios with heavy occlusions.
- Recent Advances in Embedding Methods for Multi-Object Tracking: A Survey27
This survey conducts a comprehensive overview with in-depth analysis for embedding methods in MOT from seven different perspectives, including patch-level embedding, single-frameembedding, cross-frame joint embeddedding, correlation embedding), sequential embedding; and cross-track relational embedding.
- 27
- Multi-Scale Fish Segmentation Refinement and Missing Shape Recovery27
An innovative and effective contour-based segmentation and a missing shape recovery method from an arbitrary initial segmentation that generates promising results with highly robust segmentation performance and length measurement.
- Do We Really Need a Complex Agent System? Distill Embodied Agent into a Single Model25
STEVE-2 is proposed, a hierarchical knowledge distillation framework for open-ended embodied tasks, characterized by a hierarchical system for multi-granular task division, a mirrored distillation method for parallel simulation data, and an extra expert model for bringing additional knowledge into parallel simulation data.
- 25
- Sam-Guided Enhanced Fine-Grained Encoding with Mixed Semantic Learning for Medical Image Captioning24
This paper presents a novel medical image captioning method guided by the segment anything model (SAM) to enable enhanced encoding with both general and detailed feature extraction and demonstrates the effectiveness of this approach, as it outperforms the pre-trained BLIP2 model on various evaluation metrics.
- 24
- VersaT2I: Improving Text-to-Image Models with Versatile Reward23
VersaT2I is presented, a versatile training framework that can boost the performance with multiple rewards of any T2I model using the Low-Rank Adaptation (LoRA), and introduces a gating function to combine multiple quality aspects, which can avoid conflicts between different quality aspects.
- 23
- 23
- 23
- 23
- 22
- S4Fusion: Saliency-Aware Selective State Space Model for Infrared and Visible Image Fusion21
A novel fusion framework (S4Fusion) based on the Saliency-aware Selective State Space is proposed, which introduces the Cross-Modal Spatial Awareness Module (CMSA), which is designed to simultaneously capture global spatial information from all input modalities and promote effective cross-modal interaction.
- 21
- Multi-Step Denoising Scheduled Sampling: Towards Alleviating Exposure Bias for Diffusion Models20
A multi-step denoising scheduled sampling (MDSS) strategy to alleviate the exposure bias in DDPMs and demonstrates that this approach is more effective in mitigating exposure bias in DDPM, DDIM, and DPM-solver.
- UniDCP: Unifying Multiple Medical Vision-Language Tasks via Dynamic Cross-Modal Learnable Prompts20
Unified medical vision-language model with Dynamic Cross-modal learnable Prompts, which can be plastically applied to multiple medical vision-language tasks within a unified model and conceive a dynamic cross-modal prompt optimizing strategy that optimizes the prompts within the shareable space for implicitly processing the shareable clinic knowledge.
- Self-Paced Multi-Grained Cross-Modal Interaction Modeling for Referring Expression Comprehension19
A Self-paced Multi-grained Cross-modal Interaction Modeling framework is proposed, which improves the language-to-vision localization ability through innovations in network structure and learning mechanism and significantly outperforms state-of-the-art methods on widely used datasets.
- 18
- 18
- 17
- DDMM-Synth: A Denoising Diffusion Model for Cross-modal Medical Image Synthesis with Sparse-view Measurement Embedding17
This work proposes a novel framework called the denoising diffusion model for medical image synthesis (DDMM-Synth), which combines an MRI-guided diffusion model with a new CT measurement embedding reverse sampling scheme and outperforms other state-of-the-art supervised-learning-based baselines under fair experimental conditions.
- Chasing Consistency in Text-to-3D Generation from a Single Image16
Experimental results demonstrate that Consist3D produces more consistent, faithful, and photo-realistic 3D assets compared to previous state-of-the-art methods.
- FlexiFilm: Long Video Generation with Flexible Conditions15
FlexiFilm is introduced, a new diffusion model tailored for long video generation that incorporates a temporal conditioner to establish a more consistent relationship between generation and multi-modal conditions, and a resampling strategy to tackle overexposure.
- 15
- Bionic Artificial Lateral Line Underwater Localization Based on the Neural Network Method15
A multilayer perceptron network was built in this paper to process the pressure signal and predict the coordinates on a two-dimensional plane and the experimental results show that the mutual interference of two vibration sources causes an increase in the localization error.
- ROD2021 Challenge15
The ROD2021 Challenge is the first public benchmark focusing on this topic, which attracts great attention and participation and adds strong value and a better understanding of the radar object detection task for the autonomous vehicle community.
- ASFP (Artificial Intelligence based Scoring Function Platform): a web server for the development of customized scoring functions15
An integrated platform called ASFP, a web server for the development of customized SFs for structure-based VS, and an online prediction module that provides some well-constructed target-specificSFs for VS and an additional generic SF for binding affinity prediction.
- Weakly supervised instance segmentation using multi-prior fusion14
Two types of prior knowledge, i.e., bounding box tightness prior and contour prior, are employed to guide the mask prediction and it is demonstrated that this method outperforms most state-of-the-art WSIS methods.
- 13
- 12
- Ego3DT: Tracking Every 3D Object in Ego-centric Videos11
This paper introduces a novel zero-shot approach for the 3D reconstruction and tracking of all objects from the ego-centric video, and presents Ego3DT, a novel framework that initially identifies and extracts detection and segmentation information of objects within the ego environment.
- 10
- 10
- STEVE Series: Step-by-Step Construction of Agent Systems in Minecraft9
This work constructed a hierarchical multi-agent system, augmenting it with a vision encoder and an action codebase trained on the collected high-quality dataset STEVE-21K, and enhanced it with a Critic and memory to transform it into a complex system.
- 9
- User-Aware Prefix-Tuning Is a Good Learner for Personalized Image Captioning7
A novel personalized image captioning framework that leverages user context to consider personality factors and utilizes the prefix-tuning paradigm to extract knowledge from a frozen large language model, reducing the gap between different language domains.
- 7
- Pointmap Association and Piecewise-Plane Constraint for Consistent and Compact 3D Gaussian Segmentation Field6
CCGS, a method designed to achieve both view Consistent 2D segmentation and a Compact 3D Gaussian Segmentation field, is introduced, which outperforms existing methods in both 2D panoptic segmentation and 3D Gaussian segmentation.
- 6
- 6
- VGSC-DB: an online database of voltage-gated sodium channels6
The Voltage-gated Sodium Channels Database (VGSC-DB) is the first open-source database for VGSCs, which provides open access to 6055 data records, including 3396 compounds from 173 references toward nine subtypes of Na v s.
- 6
- 6
- Unsupervised universal hierarchical multi-person 3D pose estimation for natural scenes6
An unsupervised Universal Hierarchical 3D Human Pose Estimation (UH3DHPE) method that optimizes the torso and limb poses based on a hierarchical framework is proposed and outperforms state-of-the-art methods on the multi-person moving camera scenarios.
- DAIL: Dataset-Aware and Invariant Learning for Face Recognition6
This paper proposes DAIL: Dataset-Aware and Invariant Learning, a dataset-aware loss for multi-dataset training by reducing the penalty when the same person appears in multiple datasets.
- NeuBridge: bridging quantized activations and spiking neurons for ANN-SNN conversion5
NeuBridge is introduced, an innovative ANN-SNN conversion method that utilize adaptive temporal coding to significantly reduce the number of required time steps without compromising accuracy, effectively bridging the performance gap between ANNs and SNNs.
- RIG: Synergizing Reasoning and Imagination in End-to-End Generalist Policy5
Experimental results show that the synergy of reasoning and imagination not only improves the robustness, generalization, and interoperability of generalist policy but also enables test-time scaling to enhance overall performance.
- FreeQ-Graph: Free-form Querying with Semantic Consistent Scene Graph for 3D Scene Understanding5
This paper proposes FreeQ-Graph, which enables Free-form Querying with a semantic consistent scene Graph for 3D scene understanding and designs an LLM-based reasoning algorithm that combines scene-level and object-level information to intricate reasoning.
- 5
- 5
- 5
- Handwritten Chinese signature detection with simple Copy–Paste augmentation on power plants technical documents5
This paper divides the process of signature verification into signature detection, restoration, and verification procedures to streamline automatic verification, and explores Copy–Paste augmentation to address the data scarcity problem.
- Preserve Pre-trained Knowledge: Transfer Learning With Self-Distillation For Action Recognition5
A novel transfer learning approach that combines self-distillation in fine-tuning to preserve knowledge from the pre-trained model learned from the large-scale dataset is proposed, which outperforms state-of-the-art methods on widely used UCF101 and HMDB51 datasets in action recognition task.
- Hi-LSplat: Hierarchical 3D Language Gaussian Splatting4
This paper proposes Hi-LSplat, a view-consistent Hierarchical Language Gaussian Splatting work for 3D open-vocabulary querying, which addresses the view inconsistency issue caused by 2D semantic features and introduces instance-wise and part-wise contrastive losses to capture all-sided hierarchical semantic representations.
- Distant supervised relation extraction with label entailment and collaborative denoising4
This work proposes a novel two-stage sentence-level noise reduction framework that explicitly tackles both the FN and FP problems, and proposes robust training with collaborative denoising, which dynamically removes FP noise during training by maintaining two relation classifiers simultaneously and enabling them to learn useful knowledge from each other.
- Beat the Morning Rush: Survival Analysis-Informed DNNs With Collaborative Filtering to Predict Departure Times4
A statistically-informed deep learning approach to improve commuter departure time prediction models and finds that a class of survival analysis-enhanced DNNs outperforms conventional neural networks in predicting trip departure times, while also offering more interpretability through the hazard coefficients.
- 4
- Human-Centered Prior-Guided and Task-Dependent Multi-Task Representation Learning for Action Recognition Pre-Training4
A novel action recognition pre-training framework is proposed, which exploits human-centered prior knowledge that generates more informative representation’ and avoids the conflict between multiple tasks by using task-dependent representations.
- 4
- IGFuse: Interactive 3D Gaussian Scene Reconstruction via Multi-Scans Fusion3
This work proposes IGFuse, a novel framework that reconstructs interactive Gaussian scene by fusing observations from multiple scans, where natural object rearrangement between captures reveal previously occluded regions.
- 3
- 3
- Blind Inpainting with Object-Aware Discrimination for Artificial Marker Removal3
A novel blind inpainting method that automatically reconstructs visual contents within the corrupted regions without mask input as guidance is proposed, confirming better performance over other state-of-the-art methods.
- 3
- A Class-Rebalancing Self-Training Framework for Distantly-Supervised Named Entity Recognition3
A class-rebalancing self-training framework for improving the distantly-supervised named entity recognition and a hybrid pseudo label is adopted to provide straight semantic information for the low-performance class.
- 3
- 3
- Simultaneous Intracranial Artery Tracing and Segmentation from Magnetic Resonance Angiography by Joint Optimization from Multiplanar Reformation3
The method is validated on eight TOF MRAs of both healthy subjects and patients with cerebrovascular disease, showing good agreements with human supervised tracing and segmentation results for representative features such as artery length, volume, and tortuosity.
- BioTD: An Online Database of Biotoxins2
The Biotoxins Database (BioTD) is the largest open-source database for toxins, offering open access to 14,607 data records (8,185 activity records), covering 8,975 toxins sourced from 5,220 references and patents across over 900 species.
- Cross-modal enhanced sparse CT imaging via null-space denoising diffusion with random medical measurement embedding2
The proposed Denoising Diffusion model with cross-Modal prior and physical Measurement embedding (DDMM-CT) for reconstructing sparse-view CT images demonstrates superior generalization and flexibility, allowing adjustments in the number of projections and measurement noise intensity without retraining.
- 2
- Devil in the Number: Towards Robust Multi-modality Data Filter2
The results demonstrate that the proposed text-masked filter outperforms the original CLIP score filter when selecting the top 40% of the data and the impact of numbers on CLIP and their handling provide valuable insights for improving the effectiveness of CLIP training.
- When Few-Shot Learning Meets Video Object Detection2
This paper defines the few-shot setting and creates a new benchmark dataset for few- shot video object detection derived from the widely used ImageNet VID dataset, and employs a transfer-learning framework to effectively train the video object detector on a large number of base- class objects and a few video clips of novel-class objects.
- 2
- 2
- 1
- 1
- DSG-World: Learning a 3D Gaussian World Model from Dual State Videos1
DSG-World enables efficient real-to-simulation transfer purely in the explicit Gaussian representation space, supporting high-fidelity rendering and object-level scene manipulation without relying on dense observations or multi-stage pipelines.
- 1
- –
- –
- –
- –
- –
- An Efficient Multi-prior Hybrid Approach for Consistent 3D Generation from Single Images–
Experimental results demonstrate the approach generates more consistent and realistic 3D assets than previous state-of-the-art methods in visual quality and quantitative metrics.
- –
- FrameRS: A Video Frame Compression Model Composed by Self supervised Video Frame Reconstructor and Key Frame Selector–
A self-supervised video frame reconstructor and key frame selector that can predicted the key frames with low computation costs and demonstrate computational efficiency and competitive accuracy is presented.
- Temporal Constrained Feasible Subspace Learning for Human Pose Forecasting–
This paper considers the temporal constrained feasible solutions for human pose forecasting, where the predicted poses of input historical poses are guaranteed to obey the temporal constraints strictly in the inference stage.
- –
- Handwritten Chinese Signature Detection on Scanned Technical Documents for Authenticity Verification–
- –
- –
- –
- –
- –
- A Compensatory Algorithm for High-Speed Visual Object Tracking Based on Markov Chain–
An algorithm on camera control part is proposed by utilizing a history learning model in which a one-dimension Markov Transition Matrix synthesizes some historical information efficiently to provide serials of conventional algorithms compensation when they perform worse in an interactive, high-speed circumstance, even lose the targets.
- Image Reconstruction via Small Block-Based Compressed Sensing and Discrete Legendre Polynomials–
A novel framework of compressed sensing, namely small block-based compressed sensing (SBCS) is proposed, which aims at reconstructing super resolution images built upon CS theory and the reconstruction quality is substantially superior to those of smoothing methods with filters.
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 2026-10-14. 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.