Have a look (Live Demo) of real-time face mask detection.

Datasets : https://www.kaggle.com/andrewmvd/face-mask-detection

Label_map : labelmap.pbtxt

For setup and Training:

Local/windows: https://c17hawke.github.io/tfod-setup/

For using colab gpu for training: mask_detection_ssdlie_tf1_x.ipynb
Upload all the model files with train, test record files to your drive then use colab for training
There are also another methods without uploading

For Live cam detection: mask_detect_live_cam.py

Why I used ssdlite_mobilenet_v2_coco model?

• After some investigation with the some models I decided to use ssdlite_mobilenet_v2_coco with 150k steps as it offers a faster speed and good mean_average_precision (mAP).
• I also trained with 'faster_rcnn_inception' with 200k steps and got very good accuracy but speed is very low (1 fps).

References:

https://github.com/tensorflow/models/tree/master/official
https://github.com/tensorflow/models/blob/master/research/object_detection/g3doc/tf1_detection_zoo.md
https://github.com/tensorflow/models/blob/master/research/object_detection/colab_tutorials/object_detection_tutorial.ipynb
https://tensorflow-object-detection-api-tutorial.readthedocs.io/en/latest/auto_examples/object_detection_camera.html#sphx-glr-auto-examples-object-detection-camera-py
https://c17hawke.github.io/tfod-setup/p02/