To prepare the training and test data, set model parameters and train models use the below link to run Jupyter notebook "Train_models". On the right hand side you can find a quick guide.
To use ecologically relevant metrics to test the performance of the trained model, use the below link to run Jupyter notebook "Evaluate_models". On the right hand side you can find a quick guide.
This tutorial video guides you through the process of training and testing an object detection model using the SUBSIM notebooks available on the Digital Twin Ocean infrastructure EDITO. It will introduce you to the structure of input data, explain how to train a model, inspect the results and compare these against a stronger pre-trained model. IMPORTANT: This tutorial is using a demo workflow for training, while the original SUBSIM workflows on the GitHub pages above are more elaborate.