search

lukemelas / deep-spectral-segmentation

[CVPR 2022] Deep Spectral Methods: A Surprisingly Strong Baseline for Unsupervised Semantic Segmentation and Localization
227 stars 41 forks source link

problem of dino on semantic-segmentation #6

Closed liuxingbin closed 2 years ago

liuxingbin commented 2 years ago

Hi! It's a great job for unsupervised detection and segmentation. When I try to reproduce the dino-segmentation result in table-4, I only get 19.55 which is far lower than the value reported in your paper(30.8+-2.7). Did I miss something? Looking forward to your reply.

lukemelas commented 2 years ago

Hi, thanks for the question. I received the email you sent me and responded to you there before seeing this issue. I'm closing this as a result -- if we find anything notable I'll post it here.

hq-deng commented 2 years ago

Hello, do you reproduce the result successfully? I also cannot achieve the good results.

liuxingbin commented 2 years ago

Hello, do you reproduce the result successfully? I also cannot achieve the good results.

sadly no.