Projects
Research projects and work I have been involved in.
AutoFocus-IL
2025VLM-based Visual Imitation Learning
A vision-language model guided saliency framework for data-efficient visual imitation learning. The system automatically identifies task-relevant visual cues without requiring human gaze supervision, improving policy robustness in both simulation and real-robot experiments.
Imitation LearningVLMsRobot LearningPyTorch
ORIC Benchmark
2025Object Recognition in Incongruous Contexts
A comprehensive benchmark for evaluating object recognition capabilities of large vision-language models when objects appear in unexpected or unusual settings. The benchmark tests robustness and generalization of VLMs across diverse contextual scenarios.
Computer VisionVLMsBenchmarkingEvaluation