This repository is modified from Caffe version of [yjxiong](https://github.com/yjxiong/caffe/tree/mem) and [DeepLab v2](https://bitbucket.org/aquariusjay/deeplab-public-ver2) for evaluation. We merge the batch normalization layer named 'bn_layer' in the former one into the later one while keep the original 'batch_norm_layer' in the later one unchanged for compatibility. The difference is that 'bn_layer' contains four parameters as 'slope,bias,mean,variance' while 'batch_norm_layer' contains two parameters as 'mean,variance'. Several evaluation code is borrowed from [MIT Scene Parsing](https://github.com/CSAILVision/sceneparsing).

For installation, please follow the instructions of [Caffe](https://github.com/BVLC/caffe) and [DeepLab v2](https://bitbucket.org/aquariusjay/deeplab-public-ver2). To enable cuDNN for GPU acceleration, cuDNN v5 is needed as required in 'yjxiong'. If you meet error related with 'matio', please please download and install [matio](matio) as required in 'DeepLab v2'.

The code has been tested successfully on Ubuntu 14.04 and 12.04 with CUDA 7.0.

2. Build Caffe and matcaffe:

cp Makefile.config.example Makefile.config

vim Makefile.config

3. Evaluation:

- Evaluation code is in folder 'evaluation'.

- Download trained models and put them in folder 'evaluation/model':

- pspnet50\_ADE20K.caffemodel: [GoogleDrive](https://drive.google.com/open?id=0BzaU285cX7TCN1R3QnUwQ0hoMTA)

- pspnet101\_VOC2012.caffemodel: [GoogleDrive](https://drive.google.com/open?id=0BzaU285cX7TCNVhETE5vVUdMYk0)

- pspnet101\_cityscapes.caffemodel: [GoogleDrive](https://drive.google.com/open?id=0BzaU285cX7TCT1M3TmNfNjlUeEU)

- Modify the related paths in 'eval_all.m':

- Mainly variables 'data_root' and 'eval_list', and your image list for evaluation should be similarity to that in folder 'evaluation/samplelist' if you use this evaluation code structure.

- Matlab 'parfor' evaluation is used and the default GPUs are with ID [0:3]. Modify variable 'gpu_id_array' if needed. We assume that number of images can be divided by number of GPUs; if not, you can just pad your image list or switch to single GPU evaluation by set 'gpu_id_array' be length of one, and change 'parfor' to 'for' loop.

- Run the evaluation scripts:

4. Results:

Prediction results will show in folder 'evaluation/mc_result' and the expected scores are:

(single scale testing denotes as 'ss' and multiple scale testing denotes as 'ms')

5. Demo video:

Video processed by PSPNet101 on cityscapes dataset: [Youtube](https://youtu.be/gdAVqJn_J2M)

If PSPNet is useful for your research, please consider citing:

@article{zhao2016pspnet,

author = {Hengshuang Zhao and

Jianping Shi and

Xiaojuan Qi and

Xiaogang Wang and

Jiaya Jia},

title = {Pyramid Scene Parsing Network},

journal = {CoRR},

volume = {abs/1612.01105},

year = {2016},

url = {https://arxiv.org/abs/1612.01105},

}