Released Pretrained model only get 0.714 Mean MF-ODS in Cityscapes Val Set

Lavender105 / DFF

Code for Dynamic Feature Fusion for Semantic Edge Detection https://arxiv.org/abs/1902.09104

MIT License

220 stars 51 forks source link

Released Pretrained model only get 0.714 Mean MF-ODS in Cityscapes Val Set #17

Closed convnets closed 4 years ago

convnets commented 4 years ago

I used the pretrained dff_cityscapes_resnet101.pth model and test it on Cityscapes Val Set. The performance is about 0.714 Mean MF-ODS, which is much lower than the paper reported (0.804 mean MF-ODS). The maxDist is set to 0.02 and all other settings are the same with the code you provided. Can you help explain why?

Thanks

fabriziojpiva commented 1 year ago

Hey, did you need the pretrained weights of ResNet101 (I dont mean the checkpoint of the author, but the weights of ResNet101 trained on Imagenet) to be able to reproduce 0.714 on Cityscapes? I obtain 0.27 MF-ODS