Hi Abhinav,

The drawback I’m referring to is using it in real-time scenarios(Large Scale Development).

Imagine you develop this system for the railway, then it will scan the face of every person who passes by.

The KNN model or any other model that will be trained on every person that has been missing may take time to make a prediction.

I’ve used a central database here, Firebase.

It was my college project, so I didn’t pay any attention to its infrastructure.

If the person has changed his looks then it depends on the method we are using if it is able to detect or not. Of course, there will be an effect on accuracy.