This repository it´s about Object Detection using YOLOv3 and tiny yolov3, with custom training data to detect if people is using mask or not (also I trained class label 'semi-mask'; in this case isn´t work well due to lack of images in the initial training process).
Git clone this repository
Download YOLOv3 weights here and place the file in "yolo" folder. Also if you want tiny yolo model you can download weights here and place the file in "tiny_yolo" folder.
If you have Anaconda distribution in your machine you can create virtual environment, in order to not have problems with any library dependencies (you can directly use this excellent tutorial https://uoa-eresearch.github.io/eresearch-cookbook/recipe/2014/11/20/conda/) or use next steps:
User your webcam to give a try (you can use yolov3 or tiny yolo as well!)
Result:
You can make predictions over images. Example: python prediction_image.py -y yolov3 -p 1.JPG.
After '-y' you have to specify model (yolov3 or tiny). After '-p' you have to specify path to the image, in this case we use image 1.JPG.
Result: