Published work on Novel Class Discovery (NCD) for point cloud semantic segmentation in International Journal of Computer Vision (IJCV), June 2024.
Proposed an augmentation-free unsupervised point cloud representation learning method using clustering and neural rendering (IJCV, March 2024).
Introduced the first unsupervised domain adaptation technique for point cloud segmentation via compositional semantic mixing in IEEE TPAMI (December 2023).
Authored the first comprehensive survey on Moving Camera Video Anomaly Detection (MC-VAD) in Artificial Intelligence Review (October 2023).
Developed PatchMixer, a simple yet effective architecture for 3D point cloud understanding that enhances robustness to partial inputs (Image and Vision Computing, September 2023).
Designed an attentive multimodal fusion network for optical flow estimation from RGB and depth data using self- and cross-attention (IEEE RAL, July 2023).
Contributed to retinal layer segmentation in human eyes by leveraging multi-granularity visual features.