The code trains a 6 layer deep neural network, 4 convolutional layers and 2 fully connected layers for 50 epochs with the Adam Optimizer.
This model is trained on CPU.
Training Accuracy : 99.86%
Testing Accuracy : 95.99%
Train, Test split : 600 images for X_train, 150 images for X_test
You can test out this model using your own photos. Just put in the image path in the img_path variable and run the last cell.
Author : Sammya Majumdar 12th July 2020.