instalation

	https://docs.ultralytics.com/quickstart/#install-ultralytics

preparation data custom

• download model pretraining

    https://github.com/ultralytics/ultralytics
  

• using labelImg

    $ pip install labelImg
  

• make your own dataset

• save label in folder labels

• copy classes.txt in labels to root folder

• separate result labeling in train and val

• change data_custom.yaml with your own dataset

    - datasets	[magic folder in linux]
    	- train
    		+ images
    		+ labels
    	- val
    		+ images
    		+ labels
    data_custom.yaml
    classes.txt
    yolov8m.pt
  

train:

yolo task=detect mode=train epochs=100 data=data_custom.yaml model=yolov8m.pt imgsz=640

detect

• copy best.pt from runs folder to root folder
• run image.py or video.py

look at the tutorial video

https://www.youtube.com/watch?v=gRAyOPjQ9_s

author

Danu Andrean