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
- Wenyu LiuSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works43 from public data
- Deep High-Resolution Representation Learning for Visual Recognition5,145
The superiority of the proposed HRNet in a wide range of applications, including human pose estimation, semantic segmentation, and object detection, is shown, suggesting that the HRNet is a stronger backbone for computer vision problems.
- YOLO-World: Real-Time Open-Vocabulary Object Detection1,134
YOLO-World is introduced, an innovative approach that enhances YOLO with open-vocabulary detection capabilities through vision-language modeling and pre-training on large-scale datasets and proposes a new Re-parameterizable Vision-Language Path Aggregation Network (RepVL-PAN) and region-text contrastive loss to facilitate the interaction between visual and linguistic information.
- 668
- 614
- MapTR: Structured Modeling and Learning for Online Vectorized HD Map Construction454
A unified permutation-equivalent modeling approach is proposed, i.e., modeling map element as a point set with a group of equivalent permutations, which accurately describes the shape of map element and stabilizes the learning process.
- An End-to-End Automatic Cloud Database Tuning System Using Deep Reinforcement Learning398
An end-to-end automatic CDB tuning system, CDBTune, using deep reinforcement learning (RL), which enables end- to-end learning and accelerates the convergence speed of the model and improves efficiency of online tuning.
- Boundary-Preserving Mask R-CNN261
A conceptually simple yet effective Boundary-preserving Mask R-CNN (BMask R- CNN) to leverage object boundary information to improve mask localization accuracy in instance segmentation.
- Sparse Instance Activation for Real-Time Instance Segmentation210
Owing to the simple yet effective designs with instance activation maps, SparseInst has extremely fast inference speed and achieves 40 FPS and 37.9 AP on the COCO benchmark, which significantly out-performs the counterparts in terms of speed and accuracy.
- Seedance 2.0: Advancing Video Generation for World Complexity153
Seedance 2.0 has delivered significant improvements to its foundational generation capabilities and multi-modal generation performance, bringing an enhanced creative experience for end users.
- Symphonize 3D Semantic Scene Completion with Contextual Instance Queries134
A novel paradigm termed Symphonies (Scene-from-Insts), that delves into the integration of instance queries to orchestrate 2D-to-3D reconstruction and 3D scene modeling, and fosters holistic scene comprehension by capturing context through the efficient fusion of instance queries.
- Efficient and Robust 2D-to-BEV Representation Learning via Geometry-guided Kernel Transformer109
A Geometry-guided Kernel Transformer (GKT), a novel 2D-to-BEV representation learning mechanism that leverages the geometric priors to guide the transformer to focus on discriminative regions and unfolds kernel features to generate BEV representation.
- ProRes: Exploring Degradation-aware Visual Prompt for Universal Image Restoration100
Dgradation-aware Visual Prompts are presented, which encode various types of image degradation, e.g., noise and blur, into unified visual prompts, which provide control over image processing and allow weighted combinations for customized image restoration.
- PolarDETR: Polar Parametrization for vision-based surround-view 3D detection88
A surround-view 3D DEtection TRansformer, named PolarDETR achieves promising performance-speed trade-off on different backbone configurations and ranks 1st on the leaderboard of nuScenes benchmark in terms of both 3D detection and 3D tracking.
- BoxTeacher: Exploring High-Quality Pseudo Labels for Weakly Supervised Instance Segmentation79
This work presents BoxTeacher, an efficient and end-to-end training framework for high-performance weakly supervised instance segmentation, which leverages a sophisticated teacher to generate high-quality masks as pseudo labels and proposes the noiseaware pixel loss and noise-reduced affinity loss to adaptively optimize the student with pseudo masks.
- Seedance 1.5 pro: A Native Audio-Visual Joint Generation Foundation Model73
This work introduces an acceleration framework that boosts inference speed by over 10X and distinguishes itself through precise multilingual and dialect lip-syncing, dynamic cinematic camera control, and enhanced narrative coherence, positioning it as a robust engine for professional-grade content creation.
- GaussTR: Foundation Model-Aligned Gaussian Transformer for Self-Supervised 3D Spatial Understanding57
GaussTR, a novel Gaussian-based TRansformer framework that unifies sparse 3D modeling with foundation model alignment through Gaussian representations to advance 3D spatial understanding, is introduced.
- WeakCLIP: Adapting CLIP for Weakly-Supervised Semantic Segmentation57
WeakCLIP provides an effective and parameter-efficient way to transfer CLIP knowledge to refine CAM and achieves the state-of-the-art WSSS performance on standard benchmarks.
- ControlAR: Controllable Image Generation with Autoregressive Models55
ControlAR is introduced, an efficient and effective framework for integrating spatial controls into autoregressive image generation models and surpasses previous state-of-the-art controllable diffusion models, e.g., ControlNet++.
- Perceive, Interact, Predict: Learning Dynamic and Static Clues for End-to-End Motion Prediction39
PIP is proposed, the first end-to-end Transformer-based framework which jointly and interactively performs online mapping, object detection and motion prediction, and achieves better performance than tracking-based and HD-map-based methods.
- Vision-based Uneven BEV Representation Learning with Polar Rasterization and Surface Estimation38
PolarBEV is proposed for vision-based uneven BEV representation learning, which keeps real-time inference speed on a single 2080Ti GPU, and outperforms other methods for both BEV semantic segmentation and BEV instance segmentation.
- 36
- Mask-Adapter: The Devil is in the Masks for Open-Vocabulary Segmentation29
The performance limitations of mask pooling are revealed and the proposed Mask-Adapter is introduced, a simple yet effective method to address these challenges in open-vocabulary segmentation.
- 26
- 26
- Knowledge Mining with Scene Text for Fine-Grained Recognition20
An end-to-end trainable network that mines implicit contextual knowledge behind scene text image and enhance the semantics and correlation to fine-tune the image representation is proposed.
- VMA: Divide-and-Conquer Vectorized Map Annotation System for Large-Scale Driving Scene18
A divide-and-conquer annotation scheme to solve the spatial extensibility problem of HD map generation, and abstract map elements with a variety of geometric patterns as unified point sequence representation, which can be extended to most map elements in the driving scene.
- TinyDet: Accurate Small Object Detection in Lightweight Generic Detectors16
A two-stage lightweight detection framework with extremely low computation complexity, termed as TinyDet, that enables high-resolution feature maps for dense anchoring to better cover small objects, proposes a sparsely-connected convolution for computation reduction, enhances the early stage features in the backbone, and addresses the feature misalignment problem for accurate small object detection.
- CoHD: A Counting-Aware Hierarchical Decoding Framework for Generalized Referring Expression Segmentation15
A Counting-Aware Hierarchical Decoding framework (CoHD) for GRES is proposed which outperforms state-of-the-art GRES methods by a remarkable margin and boosts multigranularity comprehension with the reciprocal benefit of the hierarchical nature.
- TinyDet: accurately detecting small objects within 1 GFLOPs15
TinyDet is a series of lightweight detectors with good performance-computation trade-offs and applicable to resource-constrained mobile or edge devices that is superior to other lightweight detectors in small object detection.
- Multimodal Mamba: Decoder-only Multimodal State Space Model via Quadratic to Linear Distillation13
This work proposes mmMamba, a framework for developing linear-complexity native multimodal state space models through progressive distillation from existing MLLMs using moderate academic computational resources, and an seeding strategy to carve Mamba from trained Transformer and a three-stage distillation recipe.
- GroundingSuite: Measuring Complex Multi-Granular Pixel Grounding12
An automated data annotation framework leveraging multiple Vision-Language Model agents and a large-scale training dataset encompassing 9.56 million diverse referring expressions and their corresponding segmentations that boosts model performance to state-of-the-art levels.
- 11
- 9
- Featurized Query R-CNN8
This work presents featurized object queries predicted by a query generation network in the well-established Faster R-CNN framework and develops a Featurized Query R-Congress, which obtains the best speed-accuracy trade-off among all R- CNN detectors, including the recent state-of-the-art Sparse R-China detector.
- AziNorm: Exploiting the Radial Symmetry of Point Cloud for Azimuth-Normalized 3D Perception7
AziNorm is proposed, which normalizes the point clouds along the radial direction and eliminates the variability brought by the difference of azimuth, and remarkably improves data efficiency and accelerates convergence, reducing the requirement of data amounts or training epochs by an order of magnitude.
- Mixture-of-Depths Attention6
This work introduces mixture-of-depths attention (MoDA), a mechanism that allows each attention head to attend to sequence KV pairs at the current layer and depth KV pairs from preceding layers, and suggests that MoDA is a promising primitive for depth scaling.
- 6
- 4
- 3
- Cross-layer attentive feature upsampling for low-latency semantic segmentation3
This work proposes a novel Guided Attentive Interpolation (GAI) method to adaptively interpolate fine-grained high-resolution features with semantic features to tackle low-latency semantic segmentation.
- Bayesian CycleGAN via Marginalizing Latent Sampling.3
The proposed Bayesian CycleGAN explores the full posteriors of Bayesian cyclic model and optimizes the model with maximum a posteriori (MAP) estimation and can achieve more stable training, superior performance and diversified images generating.
- 2
- –
Publication data from OpenAlex; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-15. 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.