AbSViT
AbSViT copied to clipboard
Unsatisfactory demo result
Thanks first for your impressive work. It's really a fascinating work that inspires me a lot. My problem is when I trial the demo.ipynb, I obtained an unsatisfactory result. The result of prior mask and feedback attention with prompt of the phone as well as the latter classification example all demonstrated meaningless picture. I would like to know where may be possibly wrong and how can I fix it. Thanks again!
Hi, sorry for the late response. Are you testing the attention on single-object image or multi-object image? The phone image is single-object while the screenshot you show here seems from the multi-object attention visualization?
I tested both on single-object image and multi-object image in the demo, and both of them showed unsatisfacory result .