Set of jupyter notebooks created to showcase the capabilities of YOLOv8 computer vision model with focus on solar panels detection:
showcase_yolo.ipynb - A brief introduction to YOLOv8 and example of detection application using one of the default models.
solar_dataset_preparation.ipynb - Downloads and prepares a solar panels dataset of very-high resolution WorldView-3 satellite data.
solar_panels_detection.ipynb - Trains a YOLOv8 model using the previous solar panels dataset and performs detection.
Create an Anaconda environment:
conda create -n solar_panels_detection-env python=3.10
conda activate solar_panels_detection-env
pip install notebook
pip install ultralytics==8.1.27
pip install scikit-learn==1.4.1.post1
or use Colab (needs Drive permissions):
solar_dataset_preparation.ipynb
A model based on YOLOv8x trained for 100 epochs is provided on request (137MB).
- YOLOv8: Jocher, G., Chaurasia, A., & Qiu, J. (2023). Ultralytics YOLO (Version 8.0.0) [Computer software]. https://github.com/ultralytics/ultralytics
- Solar Panels Dataset: https://resources.maxar.com/product-samples/15-cm-hd-and-30-cm-view-ready-solar-panels-germany
- Solar Panels Labels: Clark, C. N. (2023). Solar Panels in Satellite Imagery: Object Labels. figshare. Dataset. https://doi.org/10.6084/m9.figshare.22081091.v3