Suppose I have a lot of scans of hardcopy documents, in the form of jpegs. Some of them are potentially scans of driver's licenses or identification cards. I wonder what would be a good way to identify those scans that contain driver's license/ID cards.
One thought I had was training a model or use an existing pretrained model that can detect faces. However, if the data set I have has a lot of scans of photos of people, it would cause false positives. So I am not sure how I might approach this problem.
Any thought would be much appreciated!