open-mmlab / mmocr

OpenMMLab Text Detection, Recognition and Understanding Toolbox
https://mmocr.readthedocs.io/en/dev-1.x/
Apache License 2.0
4.27k stars 743 forks source link

Where is FCENet's Pretrained Model (ImageNet)? #932

Closed qqqqq127 closed 2 years ago

xinke-wang commented 2 years ago

Hi, the default config of FCENet uses the torchvision ResNet-50 that was pre-trained on the ImageNet for initialization, which will be automatically downloaded from the torchvision website while training started.

https://github.com/open-mmlab/mmocr/blob/main/configs/_base_/det_models/fcenet_r50_fpn.py#L10