pretrained model for resnet50 backbone

feipanir / IntraDA

Unsupervised Intra-domain Adaptation for Semantic Segmentation through Self-Supervision (CVPR 2020 Oral)

https://arxiv.org/pdf/2004.07703.pdf

MIT License

270 stars 35 forks source link

Closed zhangyuygss closed 4 years ago

zhangyuygss commented 4 years ago

Hey, I want to try your model with resnet50 as the backbone, where should I download the pretrained model? Thanks!

feipanir commented 4 years ago

Hi we currently haven't tested on ResNet50 as the backbone, rather we choose ResNet101 because it could generate pretty accurate segmentation maps during inter-DA step. If you really wanna try on ResNet50, though we could not guarantee final performance, we recommend you to look at https://pytorch.org/docs/stable/_modules/torchvision/models/segmentation/segmentation.html#fcn_resnet50

zhangyuygss commented 4 years ago

Thanks, I'm trying resnet50 to save training time. I assume the performance might drop a little, but the UDA framework will still work, am I right?

feipanir commented 4 years ago

Yes ideally it would still work.