search

w1oves / Rein

[CVPR 2024] Official implement of <Stronger, Fewer, & Superior: Harnessing Vision Foundation Models for Domain Generalized Semantic Segmentation>
https://zxwei.site/rein
GNU General Public License v3.0
215 stars 19 forks source link

Regarding Table 3: results on HGFormer #55

Closed jnikhilreddy closed 2 weeks ago

jnikhilreddy commented 1 month ago

Hi,

Thanks for the great work. Can you release the config file and pretrained models for HGFormer and HGFormer-Rein used for cityscapes to ACDC results in Table 3.

w1oves commented 2 weeks ago

Hello, thank you for your support of Rein! We did not train HGFormer ourselves but directly utilized the experimental results provided in their paper, trusting their correctness. If you require related information, please refer to the HGFormer code: https://github.com/dingjiansw101/HGFormer. Furthermore, although it might be a meaningful endeavor, we have not trained HGFormer+Rein. Should you need similar experimental results, feel free to explore our DINOV2+Reins release: https://github.com/w1oves/Rein/releases/tag/Cityscapes. Our results in Table 3 are derived from this release. Once again, thank you for your attention to Rein!