Pytorch-Uodac

Installation

Requirements

• Linux OS: Ubuntu 16.04
• Python 3.5+
• PyTorch 1.1 or higher
• CUDA 9.0 or higher
• NCCL 2
• GCC 4.9 or higher
• mmcv

Install pytorch-uodac

Clone the Pytorch-Uodac repository.

git clone https://github.com/ChenYingpeng/Pytorch-Uodac
cd Pytorch-Uodac

Install build requirements and then install pytorch-uodac. Note:(We install pycocotools via the github repo instead of pypi because the pypi version is old and not compatible with the latest numpy.)

pip install -r requirements.txt
pip install "git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI"
python3 setup.py develop

Generate train and test json data

Generate tran json data

python3 tools/data_process/generate_train_json.py --xml-dir [xxx]  --json [xxx]

Example

python3 tools/data_process/generate_train_json.py --xml-dir ../underwater/optics/data/train/box/  --json ../underwater/optics/data/train/train_data_annotations.json

Generate test json data

python3 tools/data_process/generate_test_json.py --test-image-dir [xxx]  --save-json-path [xxx]

Example

python3 tools/data_process/generate_test_json.py --test-image-dir ../underwater/optics/data/test-A-image/  --save-json-path ../underwater/optics/data/annotations/test-A-image.json

Train

./tools/dist_train.sh configs/underwater/optics/xxxx.py 2

Submit

download model link passwd : rf0n

./tools/dist_submit.sh configs/underwater/optics/xxxx.py ../underwater/optics/output/xxxx/latest.pth 2 --format_only

You could find test_A_image_submission.csv on submit/.