This convolutional neural network uses the keras and tensor flow backend. It has been trained on 8000 images of cats and dogs. Given an image of a cat or a dog, it can give an accurate prediction of what the object in the image is.
This is good simple code, thank you for uploading it!
I only changed the the steps per epoch to 8000/32 for the training and 2000/32 for the test set and saved the model that is generated.
I'd love to know what kind of accuracy you got, I got the following:
for training set loss is 0.211 and the accuracy is 0.915
for test set loss is 0.531 and the accuracy is 0.797
This is better then most machine learning code I have tested up until now.
The only test it didn't pass is when I inputted a picture of my own cat and even when inputting pictures of cats from the internet where the algorithm will nearly always say 100 % dog. So I'll have to start teach my cat some dog tricks.