ConvNeXt (CVPR'2022)
@article{liu2022convnet,
title={A ConvNet for the 2020s},
author={Liu, Zhuang and Mao, Hanzi and Wu, Chao-Yuan and Feichtenhofer, Christoph and Darrell, Trevor and Xie, Saining},
journal={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
year={2022}
}
Segmentor | Pretrain | Backbone | Crop Size | Schedule | Train/Eval Set | mIoU | Download |
---|---|---|---|---|---|---|---|
UperNet | ImageNet-1k-224x224 | ConvNeXt-T | 512x512 | LR/POLICY/BS/EPOCH: 0.0001/poly/16/130 | train/val | 46.25% | cfg | model | log |
UperNet | ImageNet-1k-224x224 | ConvNeXt-S | 512x512 | LR/POLICY/BS/EPOCH: 0.0001/poly/16/130 | train/val | 48.68% | cfg | model | log |
UperNet | ImageNet-1k-224x224 | ConvNeXt-B | 512x512 | LR/POLICY/BS/EPOCH: 0.0001/poly/16/130 | train/val | 48.97% | cfg | model | log |
UperNet | ImageNet-21k-224x224 | ConvNeXt-B-21k | 640x640 | LR/POLICY/BS/EPOCH: 0.0001/poly/16/130 | train/val | 52.71% | cfg | model | log |
UperNet | ImageNet-21k-224x224 | ConvNeXt-L-21k | 640x640 | LR/POLICY/BS/EPOCH: 0.0001/poly/16/130 | train/val | 53.41% | cfg | model | log |
UperNet | ImageNet-21k-224x224 | ConvNeXt-XL-21k | 640x640 | LR/POLICY/BS/EPOCH: 0.0001/poly/16/130 | train/val | 53.68% | cfg | model | log |
You can also download the model weights from following sources:
- BaiduNetdisk: https://pan.baidu.com/s/1gD-NJJWOtaHCtB0qHE79rA with access code s757