Implementation of Facial recognition .

[ShahkarHassan]

Implement the code of Facial recognition on your own pictures / Real time.

[ShahkarHassan]

Implemented the Code which gave Good output. This real time application is working on One shot . Got some idea of real time application. Now the models generated in last Closed issue "INITIAL CODING OF FACIAL RECOGNITION" would be used and the real time analysis would be done in coming week.

The additional features such as Attendance , Staff exclusion would be considered as side outputs in coming week while the output giving age , gender ,emotion by our own models will be the main feature.

As Mehrunisa is working on Deepface , And i am working on another code. The Better of Both would be used in our project and then we will adjust that according to our need .With appropriate camera distancing and all.

[ShahkarHassan]

Outputs accuracy of that code is listed below: Gender age: Emotion loss: Accuracy:

and Model used:

[MeharunisaAshraf]

Deepface uses one of these 4 face recognition models , Or the "Ensemble" which is average of all 4 models. I implemented all 4 models on deepface but the results were not satisfactory . The dataset is of 1.5 gb and is working as desired but when it comes to female , It doesnt work on our faces . As dataset consists of data of foreigner's faces. Either we will add our own dataset in coming week or will work on the model made by Shahkar .

The output example :

As it is image of female but giving male as output. while emotion and gender are right.

We have two options either to add dataset in this or to use another model which is editable and long . But this deepface is built in and not easy to change. All 4 models were tested and each having same problem.

[MeharunisaAshraf]

We have one more possibility of adding our own gender model in it along with the emotion and age model of Deepface . Thats the task of coming week.