Research

My research focuses on multimodal unsupervised representation learning, 3D computer vision, and natural language processing. I develop methods for learning rich representations from unlabeled data across different modalities (point clouds, images, text).

Research Areas

• 3D Point Cloud Understanding: Self-supervised and semi-supervised learning for 3D point clouds; adapting large vision-language models to 3D modalities.
• Multi-modal Learning: Contrastive learning across point cloud, image, and text representations; feed-forward 3D scene reconstruction.
• Autonomous Driving Perception: Semantic segmentation of fish-eye camera images; semi-supervised approaches for real-world deployment.

Selected Publications

See all publications →