Skip to content

wxian3/scalabel

 
 

Repository files navigation


Build Status

Scalabel (pronounced "scalable") is a versatile and scalable tool that supports all kinds of annotations needed in a driving database. It supports bounding box, semantic instance segmentation, and video tracking.

Setup

  1. Checkout the code

    git clone [email protected]:ucbdrive/scalabel.git
    cd scalabel
    
  2. Create a directory <data_dir> to store the server data:

    mkdir ../data
    

    Or, for Windows users:

    mkdir ..\data
    
  3. Launch server. There are two options, either (i) to build with Docker or (ii) to build by yourself.

    1. Build and run a Docker image from the Dockerfile.

      Build by yourself

      docker build . -t 'scalabel/server'
      

      Or

      docker pull scalabel/server
      

      After getting the docker image, you can run the server

      docker run -it -v `pwd`/../data:/data -p 8686:8686 scalabel/server
      
    2. Build the server by yourself.

      1. Install GoLang. Refer to the instruction page for details.
      2. Install GoLang dependency
      go get gopkg.in/yaml.v2
      
      1. Compile the server
      go build -i -o $GOPATH/bin/scalabel ./server/go
      

      For Windows users, use

      go build -i -o %GOPATH%\bin\scalabel.exe .\server\go
      

      Note that you may choose your own path for the server executable. We use $GOPATH/bin/scalabel as it conforms to golang best practice, but if your GOPATH is not configured, this will not work.

      1. Specify basic configurations (e.g. the port to start the server, the data directory to store the image annotations, etc) in your own config.yml. Refer to app/config/default_config.yml for the default configurations.
      2. Launch the server by running
      $GOPATH/bin/scalabel --config app/config/default_config.yml
      

      For Windows users, run:

      %GOPATH%\bin\scalabel.exe --config app\config\default_config.yml
      

      If you used a different server path when compiling, make sure to use the correct path here.

  4. Access the server through the specifed port (we use 8686 as the default port specified in the config.yml)

    http://localhost:8686
    

Usage and Demo

Please check the documentation for detailed usage instructions.

About

Quantify computer vision performance in human terms

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages

  • JavaScript 67.6%
  • HTML 15.8%
  • Go 14.1%
  • Python 1.1%
  • CSS 1.1%
  • Shell 0.3%