MdAliAhnaf/frontal_3-category_face-mask_detection

Mobile Net V2 was trained using custom and Kaggle datasets of 9.5k human face pictures with properly, poorly, and incorrectly worn masks. The model scores 95–96 percent validation after haarcascade_frontalface_alt2.xml detects the frontal face area in real time.

Jupyter NotebookMIT