Skip to content

Latest commit

 

History

History
39 lines (27 loc) · 1.08 KB

detectron2 installation and operation tutorial.md

File metadata and controls

39 lines (27 loc) · 1.08 KB
Detection2 Installation Tutorial

  • Notice：You need to install pytorch first, or you'll get an error

  • Download the archive：

git clone https://github.com/facebookresearch/detectron2.git
  • Installation：
python -m pip install -e detectron2
  • CPU running example (this is for the example, try it out)
python demo/demo.py --config-file configs/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml \
--input .\input.jpg \
--output .\output.jpg \
--opts MODEL.DEVICE cpu
Train Data

  • Run detection (with arrows)

    • Run detections (with arrow) and place the dataset in the detectron2 sibling directory (note the second detectron2 sibling directory, along with the existing output folder) :

    1671285054390

    • Change dataset name: Change the dataset to dataset_arrow, change the flow_chart_new inside to flow_chart, and put config.txt in it：

    1671285187192

    • Run the program (in the first detectron2 directory)：
    python train_keypoint.py