1 paper · 1 filter
Mazda Moayeri, Michael Rabbat, Mark Ibrahim +1
Vision-language models enable open-world classification of objects without the need for any retraining. While this zero-shot paradigm marks a significant advance, even today's best…