PotBot

YOLOv5 instruction

YOLOv5 github is added to YoloPothole folder. Installation & Setup ref: https://docs.ultralytics.com/yolov5/train_custom_data/

or: navigate to yolov5 folder, and run pip install -r requirements.txt

Ivy's personal note:

Setup:

Please run in yolov5 folder train command: python3 train.py --img 640 --epochs 3 --data coco128.yaml --weights yolov5s.pt val command: python3 val.py --weights MODEL_FILE_PATH.pt --data SunglassedWebCam.yaml --img 640 test command:

python3 detect.py --weights PATH --source --> 0 for webcam other useful features: --conf_thres = 0.25 % confidence threshold --save-text to save coordinate --name "NAME" to specify document name ---> the detect code can run on video feed, so simply add the detect function should allow us to run images through the model.

how to get box coords

See line 165 - 167 in detect.py

other

All test results and models are in ~/YoloPothole/yolove5/runs/

.yaml is in the data folder