Harvard's free programming classes teach you how to think, debug, and adapt in an AI-driven world where knowing code matters more than ever.

Abstract: As a core component of intelligent surveillance and autonomous driving systems, visual sensor-based trajectory multimodality prediction can significantly improve their perception and ...

One of the principal challenges in building VLM-powered GUI agents is visual grounding, i.e., localizing the appropriate screen region for action execution based on both the visual content and the ...