Abhishek Peri
Papers
1
Total Citations
6
H-Index
1
About
Abhishek Peri is a computer vision researcher whose work focuses on advancing local feature matching—a critical component for applications in robotics, 3D reconstruction, and visual localization. His most notable contribution, the paper "ReF -- Rotation Equivariant Features for Local Feature Matching" (2022), addresses a fundamental limitation in learning-based feature matching: the lack of inherent rotation invariance. Rather than relying solely on data augmentation, Peri introduces rotation-equivariant neural architectures that explicitly encode orientation information, enabling more robust matching under challenging viewpoint and appearance changes. This work, which has garnered 6 citations, represents an important step toward making feature matching more reliable in real-world, unconstrained environments. Peri’s research sits at the intersection of geometric deep learning and visual perception, with implications for autonomous navigation and augmented reality. By tackling the problem of equivariance in learned features, he contributes to a deeper understanding of how neural networks can incorporate geometric priors—a direction that promises to make computer vision systems more sample-efficient and robust to domain shifts.
Research Focus
Key Achievements
Top Papers
- 1ReF -- Rotation Equivariant Features for Local Feature Matching6 citations · 2022