INSTALL.md

Environment

After configuring cuda environment, you can create a conda environment to run example codes.

# create and enter into a conda environment
conda create -n orienmask python=3.7
conda activate orienmask

# install packages
pip install Cython
pip install -r requirements

# compile nms extension
python setup.py develop

Dataset

1. Download 2017 COCO dataset from the official website.

   Images
       2017 Train images [118K/18GB]
       2017 Val images [5K/1GB]
       2017 Test images [41K/6GB] (optional)
   Annotations
       2017 Train/Val annotations [241MB]
       2017 Testing Image info [1MB] (optional)
   

2. Unzip and organize the dataset as follows:

   ├── /path/to/coco
   │   ├── annotations
   │   │   ├──instances_train2017.json
   │   │   ├──instances_val2017.json
   │   │   ├──image_info_test-dev2017.json
   │   ├── train2017
   │   ├── val2017
   │   ├── test2017
   

3. Create a soft link at current directory:

   ln -s /path/to/coco coco
   

4. Prepare ground truth files and sample list files for training:

   python utils/prepare_dataset.py