Amazon Rekognition offers a wide variety of services on Computer Vision and Image Recognition.
One of it's main modules is Facial Detection.
For further details on API calls and nomenclatures, please refed to the: Documentation
First we collected images from all members of the group so we could start training our Collection
Oh, for short, from now on I'll be refering to AWS Rekognition as RKG.
We'd also like to ask for contributions with our images database so we can better measure our accuracy, if some of you would gladly take around 35-50 photos and upload them to a folder on your name it would help us to train our model and se if we are really are hitting the mark.
And don't worry, we will erase all provided data when the project ends.
Okay so now we use around 25-30 images to feed our Collection, and the other 20ish will be used to see if the API can tell me who is in the photo.
RKG takes care for us on identifying the face, isolating it, ingesting it on it's model, and with that we can automatically tag it with the User ID for identification on next steps
So, since now we already have a tagged RKG Collection we can start developing the insides of our PI to communicate with AWS service. To that we'll use Python's boto3 to create a safe bridge through AWS token authentication so we can communicate with Rekognition
For the sake of this example, we structured our "Local Path Files" (Inside PI) as such: