Xiang Li

Point2RBox-v3: Self-Bootstrapping from Point Annotations via Integrated Pseudo-Label Refinement and Utilization

WOW-Seg: A Word-free Open World Segmentation Model

MotionSight: Boosting Fine-Grained Motion Understanding in Multimodal LLMs

NAIPv2: Debiased Pairwise Learning for Efficient Paper Quality Estimation

Revisiting Data Challenges of Computational Pathology: A Pack-based Multiple Instance Learning Framework

RoRecomp: Enhancing Reasoning Efficiency via Rollout Response Recomposition in Reinforcement Learning

Representation Entanglement for Generation: Training Diffusion Transformers Is Much Easier Than You Think

OpenVid-1M: A Large-Scale High-Quality Dataset for Text-to-video Generation

See through the Dark: Learning Illumination-affined Representations for Nighttime Occupancy Prediction

Revisiting End-to-End Learning with Slide-level Supervision in Computational Pathology

SinPoint: A Novel Topological Consistent Augmentation for Point Cloud Understanding

Rethinking Point Cloud Data Augmentation: Topologically Consistent Deformation

Efficient Multi-Level Learning for Dense Object Detection