Face Recognition By Transfer Learning

In this project, I made a face recognition model by using MobileNet Architecture and through the concept of Transfer Learning. I added the dataset of two of my favourite singers,one bollywood singer "ATIF ASLAM" and one punjabi singer "PARMISH VERMA" and trained my machine to recognize their faces. For reference, you can see my project on my LinkedIn Handle.

Requirements

• Own Dataset
• Jupyter Notebook IDE
• Should know the concepts of MobileNet Architecture and Transfer Learning

Output

I used MobileNet Architecture (pre-trained model) to train my dataset and only did 2 Epochs due to which I got train_accuracy as 50% and validation_accuracy as 60%.So there are chances that may be sometimes machine predict wrong results because my accuracy is not so good. For achieving better accuracy we need large dataset and I only use less than 100 images to train my model, so if you wanna achieve better results so you should collect large dataset as well as requires more computing power.