StephanST/

pools_and_solarpanels_satellite

Model is based on Yolov8 format, specifically Yolov8l-p2.

You should run it with SAHI or some form of sliding window if your input is massive satellite images. You can use this for instance: https://huggingface.co/StephanST/WALDO30/blob/main/run_sliced_inference.py

It's trained with my own training code so not using Ultralytics, but I do recommend you use Ultralytics for inference their lib is really nice and makes things easier.

It does this:

Let me know if you use it, I could retrain it to make it a bunch better if there is any interest.