Published:
This project introduces a continual test-time adaptation method designed to handle domain shifts in real-world scenarios by adapting models on the fly without access to source data or retraining (Online Adaptation). 
Published:
We participated in the Continual Learning Challenge 2024, aiming at pushing the boundaries of continual learning techniques, addressing catastrophic forgetting, and improving model adaptation on sequential vision tasks. 
Published:
We developed a simulation pipeline for a Kinova Jaco 6‑DOF robotic arm, integrating Unity and ROS with computer vision to perform real-world tasks like leaf disease detection and garbage sorting.