Unofficial implementation of "Pyramid Scene Parsing Network" (https://arxiv.org/abs/1612.01105). This repository is just for caffe to pytorch model conversion and evaluation.

• pytorch
• click
• addict
• pydensecrf
• protobuf

Instead of building the author's caffe implementation, you can convert off-the-shelf caffemodels to pytorch models via the caffe.proto.

This step can be skipped. FYI.
Download the author's caffe.proto into the libs, not the one in the original caffe.

# For protoc command
pip install protobuf
# This generates ./caffe_pb2.py
protoc --python_out=. caffe.proto

1. Find the caffemodels on the author's page (e.g. pspnet50_ADE20K.caffemodel) and store them to the data/models/ directory.
2. Convert the caffemodels to .pth file.

python convert.py -c <PATH TO YAML>

python demo.py -c <PATH TO YAML> -i <PATH TO IMAGE>

• With a --no-cuda option, this runs on CPU.
• With a --crf option, you can perform a CRF postprocessing.

PASCAL VOC2012 only. Please set the dataset path in config/voc12.yaml.

python eval.py -c config/voc12.yaml

88.1% mIoU (SS) and 88.6% mIoU (MS) on validation set.
NOTE: 3 points lower than caffe implementation. WIP

• SS: averaged prediction with flipping (2x)
• MS: averaged prediction with multi-scaling (6x) and flipping (2x)
• Both: No CRF post-processing

• Official implementation: https://github.com/hszhao/PSPNet
• Chainer implementation: https://github.com/mitmul/chainer-pspnet