This is an AI model developed using tensorflow==2.13.0.
The model is saved as cats_and_dogs.keras
All of the code used to create the model is in cats_and_dogs.ipynb.
Model's performance: loss: 0.4881 - accuracy: 0.7674 - val_loss: 0.5053 - val_accuracy: 0.7527.
As you can see, it is not a perfect model but it can at least differentiate between dogs and cats quite a lot more than 50% of the time. It does fail sometimes though. This is largely due to lack of computing power on my part. The model has only been trained for 3 epochs, it requires a bit more epochs to be well trained. I could reduce the image size but I would prefer for the network to learn the patterns properly.
If you upload neither a dog nor a cat you get nonesensical results.
Demo:
