Projects

Ultra-high Resolution Image Understanding

• From Contexts to Locality: Ultra-high Resolution Image Segmentation via Locality-aware Contextual Correlation ICCV, 2021 [pdf][arxiv][code]

• Ultra-high Resolution Image Segmentation via Locality-aware Context Fusion and Alternating Local Enhancement IJCV, 2024 [arxiv]

• Memory-Constrained Semantic Segmentation for Ultra-High Resolution UAV Imagery IEEE RA-L, 2024 [paper][arxiv][code]

• A deep learning-based stripe self-correction method for stitched microscopic images Nature Communications, 2023 [paper][code]

• Playing to the Strengths of High- and Low-Resolution Cues for Ultra-high Resolution Image Segmentation IEEE RA-L, 2025 [paper][code]

Multimodal Semantic Segmentation

• Keep the Balance: A Parameter-Efficient Symmetrical Framework for RGBX Semantic Segmentation CVPR 2025 [paper] [code]
  • Summary: Introduces a parameter-efficient symmetrical framework for RGB-X semantic segmentation (X = depth/thermal/NIR).
  • Core innovations:
    • Symmetrical modality interaction via shared feature rectification.
    • Lightweight fusion with dynamic token selection.
    • ~40% parameter reduction while maintaining accuracy.
    • Achieves +2.1% mIoU on NYUDv2 (RGB-D) and +1.8% mIoU on Cityscapes (RGB-T).

BEV Segmentation