Learning to Exploit the Prior Network Knowledge for Weakly-Supervised Semantic Segmentation

By Carolina Redondo-Cabrera, Marcos Baptista-Ríos and Roberto J. López-Sastre.

This is a repository with the original implementation of the weakly-supervised semantic segmentation method detailed in our IEEE TIP journal paper.

Example of our semantic segmentation results.

License

This repository is released under the GNU General Public License v3.0 License (refer to the LICENSE file for details).

Citing

If you make use of this data and software, please cite the following references in any publications:

@Article{Redondo-Cabrera2019,
author 	= {Redondo-Cabrera, C. and Baptista-R\'ios, M. and L\'opez-Sastre, R.~J.},
title   = {Learning to Exploit the Prior Network Knowledge for Weakly-Supervised Semantic Segmentation},
journal = {IEEE Transactions on Image Processing},
year    = {2019},
}

@article{Deeplab,
title={DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs},
  author={Liang-Chieh Chen and George Papandreou and Iasonas Kokkinos and Kevin Murphy and Alan L Yuille},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year={2018}
}

Contents

1. Requirements: software
2. Requirements: hardware
3. Installation
4. Pre-trained models
5. Usage

Requirements: software

1. Requirements for Caffe and pycaffe (see: Caffe installation instructions)
2. Matlab 2015 (or higher)
3. A copy of the PASCAL VOC 2012 Development Kit

This project has been developed and tested under Ubuntu 14.04 and Ubuntu 16.04.

Requirements: hardware

1. It is required a GPU with at least 11 GB of memory and CUDA support. We have used an NVIDIA Titan and an NVIDIA Geforce 1080 Ti.

Usage

Step 1: Clone the repository

git clone https://github.com/gramuah/weakly-supervised-segmentation.git

Step 2: We'll call the directory that you cloned PROJECT_ROOT

Step 3: Build the Caffe distribution we provide.

```Shell
cd $PROJECT_ROOT/caffe
# Now follow the Caffe installation instructions here:
#   http://caffe.berkeleyvision.org/installation.html

# If you're experienced with Caffe and have all of the requirements installed
# and your Makefile.config in place, then simply do:
make -j8 && make pycaffe

```

Step 4: Run the following script to generate the pre-trained models provided

cd $PROJECT_ROOT/caffe
generate_caffe_models.sh

Check that the models downloaded have the correct MD5 signature.

Step 5: You are ready to run the following script to generate the semantic segmentation.

cd $PROJECT_ROOT/caffe
python run_semantic_segmentation.py

Note that we provide 2 examples images in folder:

$PROJECT_ROOT/datasets/benchmark_RELEASE/dataset/JPEGImages/

You can add more images to this directory. If so, remember to also update the files: caffe/voc12_weakly/list/val.txt caffe/voc12_weakly/list/val_id.txt

Output is written in $PROJECT_ROOT/voc12_weakly/res/.

Enjoy.