Rishabh Kabra
Papers
1
Total Citations
13
H-Index
1
About
Rishabh Kabra is a researcher advancing unsupervised visual learning, with a focus on object-centric representation and scene decomposition. His most-cited work, "PARTS: Unsupervised segmentation with slots, attention and independence maximization" (2021, 13 citations), introduces a novel framework that combines slot-based attention with independence maximization to segment visual scenes into coherent objects without labeled data. This approach tackles a fundamental challenge in computer vision: enabling models to perceive the world as humans do—as composed of distinct objects with properties like shape, size, and color—purely from raw pixels. By leveraging attention mechanisms and promoting independence among learned representations, PARTS achieves robust unsupervised segmentation, offering a pathway toward more interpretable and compositional AI systems. Kabra’s contributions sit at the intersection of cognitive science and deep learning, aiming to bridge the gap between human perception and machine understanding. His work has influenced subsequent research in object-centric learning and generative models, demonstrating impact through citations and inspiring further exploration into how machines can autonomously discover structure in complex visual data.
Research Focus
Key Achievements
Top Papers
- 1