This repo holds the model for the Masuku Project
The model is based on a pre-trained YOLOv5 Image Classification Model We used a model that was trained on a dataset of surgical masks and then applied transfer learning using our dataset Our dataset contained images of people wearing various forms of face coverings like helmets, masks, bandanas, ski masks and so on
Firstly, we used the yolo_test.ipynb to train the image on our dataset. The model was trained for 50 epochs and came out with an accuracy of 95%
We exported the model as model.pt and model.onnx files. We decided to use model.onnx because of the ease of work
- Pass an image as input
- The output returns a matrix of shape (1, 1, 25200, 7)
- From that matrix we were able to extract the coordinates, width and height of the bounding box and the confidence of the model
- Then we use OpenCV to draw the bounding box on the image
Now, this is not how the final version of the program will work. We plan to use live footage of a person and take the first frame of their appearance as the input image Then we can say whether the person can use the ATM or No