Face Recognition with Transfer Learning

Transfer Learning: Transfer learning is a machine learning model developed for a task that is reused as a starting point for a different task. We can give a new dataset to fine-tune the pre-trained CNN. If the new dataset is small, it's better to add some layers before the output layer at the end and retrain new layers.

Face Recognition: It's a technology to identify/verify a person from a digital image or video.

In this project work, I used the concept of transfer learning with the VGG16 pre-trained model for face recognition of the trained faces. In this small model, I have used only 2 classes of face for training. One is my own face and the other is Bollywood star, Hrithik Roshan. For this, I provided some training to VGG16 (pre-trained model) with 'imagenet' images.

Here are some project details:

• Import VGG 16 model with dataset.

• Freeze the pre-trained layers.

• Add and train new layers.

• Fit the model and got an accuracy of around 96%.

• Finally, test the model.

• Found major weight for the first value in the array that is the correct result.