Using Yolo, OAK-D and Python to train and run a system for FOD detection in a aerospace setting.
The repository contains both code to train a model in ONNX format and convert it to BLOB format for run time use on an OAK-D device. All the training is in yolov8_pascal_voc_training.py
The classify_X files are for real-time detection of objects from raw images or from an attached OAK-D camera.
The openvino_model directory contains a trained Yolov8n model trained using the Omaha FOD dataset https://github.com/FOD-UNOmaha/FOD-data. The model using Yolov8 with 31 classes using 300x300 images.
Enjoy