Jawabreh0
commented
10 months ago first of all if your working with real life scenario, you have to think how you're going to get the training data, because you'll need huge amount of samples for the same face to get high accuracy. in my case we made a desktop app linked to professional camera that takes a full face scan of the individuals, this face scan is 5-15 seconds video record, then another algorithm is taking this video scan and dividing it to 1000 image then you can use this images for the training process. check the below repo to get the algorithm that's diving the face scan video. https://github.com/Jawabreh0/Divide-Video
Anyway, the direct answer for your question is to get your data, divide it to training, validation, and testing data, save the training data in a folder and name it for example "training_data", then go to get-embeddings.py file, line number 82 and change the directory location to yours. and change the location of the output in your own location (line numebr 85 and 103), at the end change the directory location at line number 117 where you will be saving the output embeddings (features).
after that take this embeddings file as input to the face recognition algorithm in the identify.py file, but you have to change the input embeddings location at line number 16, and the input test image at line number 90.
Good Luck :)
Can you Explain please, how can i use this Face recognition system for my own dataset. (I am new to this) I mean, i am woring on creating a Face Recognition based attendance system. I want help with "how can i do this using your repo?" Thanks in Advance:)