Skip to content

An image labeling tool which supports bounding boxes and segmentation masks labeling

Notifications You must be signed in to change notification settings

shiannn/Image-Labeling-Tool

Repository files navigation

A Simple Image Labeling Tool

  1. Installation
    • Install Opencv
    • Add the Absolute path to OpenCV folder containing OpenCVConfig.cmake to config_vars.txt
    • Cmake
      # In the project directory
      mkdir build/
      cd build/
      cmake ../
    • Build
      # In build/
      make
  2. Run
    • Select the directory containing images to be labeled

      # In build/
      ./DisplayImage "<Directory Path containing images>/*.jpg"
    • Select Bounding Boxes or Segmentation Masks labeling following the program

    • Select the path for saving segmentation masks. The directory will be created if not existed.

Bounding Boxes Labeling

  • Press "a" & "d" for the previous and next image

  • Left button on Mouse for adding a point of bounding-box

  • Right button on Mouse for undoing

    You will get a label.csv containing the coordinates. The $i$'th bounding box is composed of two points $(x_i^1, y_i^1)$, $(x_i^2, y_i^2)$.

    image name $x_1^1$ $y_1^1$ $x_1^2$ $y_1^2$ $x_2^1$ $y_2^1$ ...
    arrow.jpg 293 189 485 479 496 328 628
    text.png 40 71 82 122 91 70 130
    triangular_sign.png 224 485 322 587 327 481 415
  • Press "q" for quiting the program

Semgentation Masks Labeling

  • Press "a" & "d" for the previous and next image

  • Press "s" & "w" for the previous and next label index

  • Press "1" & "2" for the thinner and thicker brush

  • Left button on Mouse for labeling a segmentation mask. Drag the mouse for larger area.

  • Right button on Mouse for undoing

    You will get segmentation mask results like the following.

  • Press "q" for quiting the program

About

An image labeling tool which supports bounding boxes and segmentation masks labeling

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published