https://github.com/lolipopshock/detectron2_al

Active Learning for Object Detection using Detectron2
https://github.com/lolipopshock/detectron2_al

Last synced: 2 months ago
JSON representation

Active Learning for Object Detection using Detectron2

• Host: GitHub
• URL: https://github.com/lolipopshock/detectron2_al
• Owner: lolipopshock
• Created: 2020-07-21T05:17:45.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2020-09-30T19:05:12.000Z (about 4 years ago)
• Last Synced: 2024-04-13T09:11:47.335Z (9 months ago)
• Language: Python
• Size: 233 KB
• Stars: 44
• Watchers: 5
• Forks: 12
• Open Issues: 3
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Detectron2 for Active Learning in Object Detection 

## Usage 

1. Clone the repository with all the submodules:

    ```bash

    git clone --recurse-submodules [email protected]:lolipopshock/Detectron2_AL.git

    ```

2. Install dependencies:

    1. Installing object detection environment with according to your environment

        - The tested version of pytorch is 1.4.0 with CUDA 10

        - And you **must** install Detectron2 with version 0.1.1. Newer versions has different APIs. 

            ```bash

            pip install detectron2==0.1.1 \

                -f https://dl.fbaipublicfiles.com/detectron2/wheels/cu100/torch1.4/index.html

            ```

    2. Installing other necessary dependencies:

        ```bash

            pip install -r requirements.txt

        ```

    3. Installing UI components

        ```bash

        cd src/label-studio

        pip install -e .

        ``` 

3. Setting up the label-studio server and modeling backend

   1. Initialize the labeling server (If your image folder is `./data`)

        ```bash

        label-studio init labeling/tk-labeling \

                --input-path=./data \

                --input-format=image-dir \

                --allow-serving-local-files --force \

                --label-config=extra/config.xml \

                --ml-backends http://localhost:9090

        ```

        And you can start the server via

        ```

        label-studio start labeling/tk-labeling

        ```

    1. Initialize the model backend server

        ```bash

        label-studio-ml init labeling/backend_model --script extra/backend_model.py

        ```

        And similarly, you can start the backend server by

        ```bash

        label-studio-ml start labeling/backend_model 

        # There's a relative import of the libraries

        # So you have to run this command in the project project

        # root path to avoid import errors

        ```

4. Start using active learning for annotation