Attention-based semantic segmentation model
* You can use ./Dockerfile to build an image.
Create a directory where you can keep large files. Ideally, not in this directory.
> mkdir <large_asset_dir>
This will just print out the command but not run. It's a good way to inspect the commandline.
This will dump network output and composited images from running evaluation with the Cityscapes validation set.
You should end up seeing images that look like the following:
Train cityscapes, using HRNet + OCR + multi-scale attention with fine data and mapillary-pretrained model
The first time this command is run, a centroid file has to be built for the dataset. It'll take about 10 minutes. The centroid file is used during training to know how to sample from the dataset in a class-uniform way.