UWSegFormer

More information will be available when the paper is published

Requirements

• Python 3.6+
• Pytorch 1.3+
• mmcv-full>=1.3.17, <1.6.0 (we use mmcv 1.5.3 and mmsegmentation 0.11.0 in code)

Dataset

Please create a data folder in your working directory and put the SUIM dataset or DUT dataset in it for testing.

data
  ├── SUIM
  |   ├── annotations
  │   │   ├── trainging
  |   |   │   ├── ......
  │   │   ├── validation
  |   |   │   ├── ......
  │   ├── images
  │   │   ├── trainging
  |   |   │   ├── ......
  │   │   ├── validation
  |   |   │   ├── ......

Test

Download trained weights. Evaluate UWSegFormer on SUIM:
python tools/test.py local_configs/uwsegformer/uw.b0.640x480.suim.160k.py model_checkpoint_path --eval mIoU
or evaluate UWSegFormer on DUT:
python tools/test.py local_configs/uwsegformer/uw.b0.640x480.dut.40k.py model_checkpoint_path --eval mIoU
Note that all checkpoints and datasets are available at Google Drive