Kamalesh Palanisamy

Papers

Total Citations

H-Index

About

Kamalesh Palanisamy is a researcher at the forefront of computer vision and robotics, with key contributions in self-supervised learning, few-shot learning, and robotic manipulation. His work on **Self-Supervised Unseen Object Instance Segmentation via Long-Term Robot Interaction** (2023, 10 citations) introduces a novel robotic system that improves segmentation of novel objects by enabling robots to interact with them over extended periods—grasping, pushing, and observing—to iteratively refine segmentation masks without human labels. This advances autonomous robotic perception in unstructured environments. Palanisamy is also the lead author of **Proto-CLIP: Vision-Language Prototypical Network for Few-Shot Learning** (2023–2024, 18 citations combined), a pioneering framework that extends prototypical networks to vision-language models like CLIP. By leveraging both image and text prototypes, Proto-CLIP achieves state-of-the-art few-shot classification, enabling models to learn new concepts from just a handful of examples. His work bridges the gap between large-scale pretraining and sample-efficient adaptation, with significant implications for robotics, object recognition, and interactive AI. Palanisamy’s research demonstrates a rare ability to integrate real-world robotic interaction with cutting-edge vision-language learning, making him a rising figure in embodied AI.