1 paper · 1 filter
Qianhui Wu, Kanzhi Cheng, Rui Yang +15
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…