Papers
5
Total Citations
90
H-Index
4
About
Jungong Han is a leading researcher in computer vision and 3D perception, with a focus on RGB-D sensing, point cloud analysis, and multimodal learning. His pioneering work on Kinect-based RGB-D sensors helped democratize high-resolution depth sensing, enabling new algorithms and applications that have garnered over 55 citations. Han’s major contributions include developing MAPLE, a masked pseudo-labeling autoencoder for semi-supervised point cloud action recognition, which addresses the critical need for efficient human action recognition in autonomous driving and robotics. He also advanced incremental instance-oriented 3D semantic mapping for unknown indoor scenes, enhancing robot-environment interaction. His dual-resolution dual-path convolutional neural networks for fast object detection improved speed-accuracy trade-offs in robotic and mobile vision. Most recently, Han introduced LLMI3D, a multimodal large language model-based approach for 3D perception from a single 2D image, pushing the boundaries of generalization in open-world scenarios. With over 90 total citations and a trajectory of impactful innovations, Han’s work bridges foundational RGB-D research with cutting-edge AI, making him a key figure in advancing 3D understanding for real-world applications.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5LLMI3D: MLLM-based 3D Perception from a Single 2D Image2 citations · 2024