KristenKehrer/schoolbus_yolov5

schoolbus_yolov5

This project uses yolov5 to detect the school bus passing my house and then send a text.
I'll be linking blog article tutorials to walk you through the project here when they're completed.

See the training metrics here: Workbook

Running

1. create virtual env directory venv/ with pycharm or from the command line

    python -m venv venv
   

2. install everything with install.ps1

    .\install.ps1
   

3. create comet artifact by running create_comet_artifact.py

    python create_comet_artifact.py
   

4. run training with train.ps1

    .\train.ps1
   

5. run detection program detect_bus.py:

    python detect_bus.py
   

Full list of documentation

• aws.py - Utility functions that use AWS
• camera.py - Functions for connecting to the webcam
• check_cuda.py - Super short file to check that CUDA is working
• create_comet_artifact.py - create a comet artifact from a roboflow export
• detect_bus.py - the bus detector script!
• detector.py - A class for managing notifications
• install.ps1 - The initial install script for setting up dependencies
• setup_credentials.py - writes all the needed credential files for authentication with stuff
• train.ps1 - script to run yolov5 training on custom data set
• yolov5_detect.py - my copy of the detect.py script from yolov5