Open wibimaster opened 3 years ago
I checked how it works in "test_image.py".
1] But it seems that we have to use one image for one label ; so, we cannot add multiple photo for one person, to increase probabilities to have a good result ? Like it's done on CASIA data set for example
2] Can you correct me if I'm wrong ; here you didn't do transfer learning with new faces/labels on the existing trained CASIA dataset ? I'm not sure about the way you do face recognition (not detection) with or without the CASIA...
Thanks !
Hi !
The README lacks some documentation to know how to add (re-train) with some new faces ; can you please add something ?
I think there's 2 points here :
I understand there is a database-like to manage this, is there a way to manage the database content itself ? (add/removes faces and labels...)
Thanks !