How to do Object Recognition with TensorFlow(Keras) the Easiest way

We can see that we have more than 20 Millions parameters, that only 132,613 of them are trainable, and all are in the last two layers. Finally, we fit the train data to the model and set the batch_size, epochs, and validation data to test the model's performance on the test set. NOTICE: usually, 32 is the best choice for batch_size, but if you want to change it, it is efficient to use the numbers produced by this formula: 2 n.