Skip to content
Branch: master
Find file Copy path
Find file Copy path
3 contributors

Users who have contributed to this file

@ppwwyyxx @wanyenlo @facebook-github-bot
80 lines (60 sloc) 3.2 KB

Getting Started with Detectron2

This document provides a brief intro of the usage of builtin command-line tools in detectron2.

For a tutorial that involves actual coding with the API, see our Colab Notebook which covers how to run inference with an existing model, and how to train a builtin model on a custom dataset.

For more advanced tutorials, refer to our documentation.

Inference with Pre-trained Models

  1. Pick a model and its config file from model zoo, for example, mask_rcnn_R_50_FPN_3x.yaml.
  2. Run the demo with
python demo/ --config-file configs/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x.yaml \
  --input input1.jpg input2.jpg \
  --opts MODEL.WEIGHTS detectron2://COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_3x/137849600/model_final_f10217.pkl

The configs are made for training, therefore we need to specify MODEL.WEIGHTS to a model from model zoo for evaluation. This command will run the inference and show visualizations in an OpenCV window.

For details of the command line arguments, see -h. Some common ones are:

  • To run on your webcam, replace --input files with --webcam.
  • To run on a video, replace --input files with --video-input video.mp4.
  • To run on cpu, add MODEL.DEVICE cpu after --opts.
  • To save outputs to a directory (for images) or a file (for webcam or video), use --output.

Use Detectron2 in Command Line

We provide a script in "tools/", that is made to train all the configs provided in detectron2. You may want to use it as a reference to write your own training script for a new research.

To train a model with "", first setup the corresponding datasets following datasets/, then run:

python tools/ --num-gpus 8 \
	--config-file configs/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x.yaml

The configs are made for 8-GPU training. To train on 1 GPU, change the batch size with:

python tools/ \
	--config-file configs/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x.yaml \

For most models, CPU training is not supported.

(Note that we applied the linear learning rate scaling rule when changing the batch size.)

To evaluate this model's performance, use

python tools/ \
	--config-file configs/COCO-InstanceSegmentation/mask_rcnn_R_50_FPN_1x.yaml \
	--eval-only MODEL.WEIGHTS /path/to/checkpoint_file

For more options, see python tools/ -h.

Use Detectron2 in Your Code

See our Colab Notebook to learn how to use detectron2 APIs to:

  1. run inference with an existing model
  2. train a builtin model on a custom dataset

See detectron2/projects for more ways to build your project on detectron2.

You can’t perform that action at this time.