Papers
1
Total Citations
7
H-Index
1
About
Tianhao Gu is a researcher advancing the frontiers of computer vision and robotics, with a primary focus on unsupervised depth prediction and camera pose estimation from unlabeled video data. His most cited work, "OnionNet: Single-View Depth Prediction and Camera Pose Estimation for Unlabeled Video" (2020, 7 citations), introduces a novel framework that enables machines to perceive spatial relationships without human supervision—mimicking the innate visual ability of humans to infer distance and position. The OnionNet architecture, comprising LeafNet and ParachuteNet, represents a significant contribution to self-supervised learning, allowing robots to understand their environment from raw video streams alone. This work addresses a critical challenge in autonomous systems: reducing reliance on expensive labeled datasets while maintaining robust 3D perception. Gu's research sits at the intersection of geometric deep learning and robotic perception, with implications for autonomous navigation, augmented reality, and scene understanding. By developing methods that learn from unlabeled data, he is helping to democratize advanced visual capabilities for robots, making them more adaptable and cost-effective in real-world applications. His contributions continue to influence the growing field of unsupervised visual learning.
Research Focus
Key Achievements
Top Papers
- 1