SELMA3D 2025: Example for building a container image of your algorithm for Preliminary Development Phase Task 1 with SSL

Step 1: Implement your solution

• In requirements.txt, list the packages required for your solution.

• In inference.py, load_image_file_as_array function will automatically load the testing image once you submit your algorithm container. Do not change the image reading and saving parts.

• Modify the processing part in inference.py file to preprocess the input image and define the model.

• Put the resources required for prediction to the resource folder such as model checkpoints, then modify the resource part in inference.py file to load the resources.

• Modify the prediction part in inference.py file, replacing it with your solution to make a prediction for the loaded image array.

Step 2: Test locally

• Call the do test_run bash script using the command: ./do_test_run.sh

  This will start the inference and reads from /test/input and outputs to /test/output.

Step 3: Save the container image for Grand Challenge submission

• Call the do save bash script using the command: ./do_save.sh

  This will create a container image of the algorithm.

Step 4 : Use the github repo for Grand Challenge submission (optional)

• Fork this repository to a new repository under your GitHub account.
• In your repository, complete Step 1 and Step 2 to implement and test your solution.
• Follow the instructions below for submitting to the Grand Challenge (you may need to copy and paste the link into your browser to access it): https://grand-challenge.org/documentation/linking-a-github-repository-to-your-algorithm/