RT-VLM: Re-Thinking Vision Language Model with 4-Clues for Real-World Object Recognition Robustness

Junghyun Park, Tuan Anh Nguyen, Dugki Min

Real world deployments often expose modern object recognition models to domain shifts that precipitate a severe drop in accuracy. Such shifts encompass (i) variations in low level…