MaskFormer (NeurIPS'2021)
@inproceedings{cheng2021per,
title={Per-pixel classification is not all you need for semantic segmentation},
author={Cheng, Bowen and Schwing, Alex and Kirillov, Alexander},
booktitle={Thirty-Fifth Conference on Neural Information Processing Systems},
year={2021}
}| Segmentor | Pretrain | Backbone | Crop Size | Schedule | Train/Eval Set | mIoU | Download |
|---|---|---|---|---|---|---|---|
| MaskFormer | ImageNet-1k-224x224 | Swin-Tiny | 512x512 | LR/POLICY/BS/EPOCH: 0.00006/poly/16/130 | train/val | 47.31% | cfg | model | log |
| MaskFormer | ImageNet-1k-224x224 | Swin-Small | 512x512 | LR/POLICY/BS/EPOCH: 0.00006/poly/16/130 | train/val | 49.91% | cfg | model | log |
| MaskFormer | ImageNet-22k-384x384 | Swin-Base | 640x640 | LR/POLICY/BS/EPOCH: 0.00006/poly/16/130 | train/val | 53.22% | 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