Madhu Vankadari
Papers
1
Total Citations
7
H-Index
1
About
Madhu Vankadari is a researcher whose work lies at the intersection of computer vision and deep learning, with a particular focus on 3D scene understanding and autonomous perception. His most notable contribution is an unsupervised deep learning framework that leverages Bayesian inference to predict both depth and confidence maps from monocular RGB images. This innovative approach, published in 2020, addresses a critical challenge in self-supervised depth estimation: quantifying prediction uncertainty. By simultaneously learning to predict per-pixel depth, camera pose, and a confidence map, Vankadari’s method enables more reliable scene reconstruction without requiring ground-truth depth data. The framework has garnered 7 citations, reflecting its relevance to researchers working on monocular depth estimation and uncertainty-aware learning. This work is particularly impactful for applications in autonomous driving, robotics, and augmented reality, where accurate and trustworthy depth perception is essential. Vankadari’s research continues to push the boundaries of unsupervised learning for geometric computer vision, offering elegant solutions that bridge probabilistic reasoning with modern deep learning architectures.
Research Focus
Key Achievements
Top Papers
- 1