This repo contrains files needed to get YOLO running on a dataset quickly and easily.
To use, you need a dataset of images, normally ~100 is a good start but this depends on the specific use case and precision needed.
All files were made with google colab in mind but it should be fine to run them on any jupyter environemnt. The annotation widget requires a jupyter notebook environment to run but the other scripts could easily be adapted for any python environment
- annotate.ipynb: This file runs the annotation widget with which you can draw your training bounding boxes, and converts them into a format that YOLO can use for training
- train.ipynb: This trains the YOLO model using the output from the annotate file
- run.ipynb: This gives a basic example of how to use a trained model and interact with the results in python. You will likely want to adapt this for your use case.
Additional running instructions are included in the files themselves