Safety hat and vest detection with Yolo v4

Question

I am training a Yolo v4 object detection model that detects the wearing of safety hats and vest on construction sites, I want to show as a result if the person is wearing the hat and vest or not, The bounding box of the person must show "person wearing hat and vest" or "person not wearing hat and vest" depending on the results of the bounding boxes of the objects:person, hat, vest. Is there any idea on how I can do it ? how can i find the distance or make a relationship between the bounding boxes? or at least how can i access to the bounding boxes coordinates to use it to code.example of detection

Answer 1

How did I do it?

1. Take the data and figure out what transfer learning model would work best?----->I chose YOLO v4
2. Convert the provided dataset into DarkNet dataset format for training on the YOLO v4 model @Roboflow.com
3. Training the YOLO v4 model on my new custom data set on Google colab (Why not local system? Because i needed to use the GPU for faster training of the model)
4. Running the video Frame by Frame and using the model to make bounding boxes around the face and labelling it.
5. Saving the labelled pictures in a video format.