/image-dataset-augmentation

Be able to generate an image dataset with a desired image overlaid on top, everything from a desired region of interest extracted from an input image.

Primary LanguagePythonMIT LicenseMIT

Image Data Augmentation

Goal

Be able to generate an image dataset with a desired image overlaid on top.

How to run

  • Move into the project root file

  • Install base Python dependencies

    • pip install -r etc/local/requirements.txt

  • Set the paths for the INPUT and BACKGROUND images, DESTINATION directory and X & Y Axis iterations

    • Set ENVIRONMENT VARIABLES for the paths to the images, destination dirs, iterations

      • export LOCAL_IMAGE_INPUT_SOURCE=${IMAGE_PATH}
      • export LOCAL_IMAGE_BACKGROUND=${IMAGE_PATH}
      • export LOCAL_DESTINATION_IMAGE_DIRECTORY=${DESTINATION_DIR_PATH}
      • export LOCAL_X_AXIS_ITERATIONS=${X_AXIS_ITERATIONS}
      • export LOCAL_Y_AXIS_ITERATIONS=${Y_AXIS_ITERATIONS}

    • For default values, modify src/settings/local.py to have your desired input image path and background image path

      • LOCAL_IMAGE_INPUT_SOURCE
      • LOCAL_IMAGE_BACKGROUND
      • LOCAL_DESTINATION_IMAGE_DIRECTORY
      • LOCAL_X_AXIS_ITERATIONS
      • LOCAL_Y_AXIS_ITERATIONS

  • Run the project applicaton

    • python src/app.py