This includes all of the files from my Saturn Cloud YOLOv8 environment. To run, make sure your data is stored in datasets/data in their appropriate folders. YOLO is very particular about their names and where to store the files, so try not to change the configuration if you want to do so. To run these, first put your test and train data in their respective folders. Make sure your images are in the image folder and then your masks are in the labels folder. Then you can run preprocessing.ipynb which will automatically 1) resize your images for the 640x640 pixel requirement in YOLO, 2) generate bounding boxes from the masks and replace the masks .png files with the bounding box .txt files, and 3) move 20% of the images and labels from the train into the validation folder.

This is the configuration in which YOLO reads to understand where the data is and what are classes that it is segmenting

This is the notebook transfer data from UNET type to YOLO type.

This is the notebook transfer data from YOLO type to UNET type.

The simple notebook that runs the model. Currently experiencing issues relating to the CUDA going outside of the bounds.

A pre-trained model from Ultralytics that does segmentation.