By Wei Liu, Andrew Rabinovich, and Alex Berg.

ParseNet is a unified framework for semantic segmentation with CNN. You can use the package to train/evaluate a network for segmentation. For more details, please refer to our arXiv paper.

Please cite ParseNet in your publications if it helps your research:

@article{liu15parsenet,
  Author = {Liu, Wei and Rabinovich, Andrew and Berg, Alexander},
  Journal = {arXiv preprint arXiv:1506.04579},
  Title = {ParseNet: Looking Wider to See Better},
  Year = {2015}
}

1. Installation
2. Model

1. Get the code

git clone https://github.com/weiliu89/caffe.git
cd caffe
git checkout fcn

2. Build the code. Please follow Caffe instruction to install all necessary packages and build it.

make -j8
make mat
make py
make test -j8
make runtest -j8

Note: Since ParseNet merges #2016, which will cause a crash on exit. You can safely ignore it as it is a known side effect.

syncedmem.cpp:16] Check failed: error == cudaSuccess (29 vs. 0) driver shutting down

The ParseNet model and solver on PASCAL has been shared at the Caffe's Model Zoo. Please check it out for more details.