Ubuntu18.04 configure detectron2 to achieve Faster RCNN target detection and Mask RCNN instance segmentation (1)

1. Background introduction

• Detectron2 project address: https://github.com/facebookresearch/detectron2
• FasterRCNN paper: https://arxiv.org/abs/1506.01497
• COCO data set: http://cocodataset.org/#home
• VOC data set: http://host.robots.ox.ac.uk/pascal/VOC/

2. Environmental requirements

• Ubuntu 18.04
• CUDA10
• pytorch==1.3.0
• torchvision==0.4.1
• pycocotools

3. Installation steps

3.1 Installation dependencies

• cocoapi

git clone https://github.com/cocodataset/cocoapi.git
cd cocoapi/PythonAPI
python setup.py build_ext install

• Cython

pip install cython

3.2 Install detectron2

git clone https://github.com/facebookresearch/detectron2.git
cd detectron2 && python -m pip install -e .

4. Test the installation

After the installation is complete, you can enter the following command to test whether the installation is successful.

python3 demo.py --config-file ../configs/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml \
  --input cat.jpg \
  --output result_cat.jpg \
  --opts MODEL.WEIGHTS detectron2://COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x/137849600/model_final_f10217.pkl

The input picture is: the

result picture is: the

test is successful, indicating that the installation is successful, and the framework can be used for further training, testing, etc.