Classifying cats and dogs using CNNs

I made a Convolutional Neural Network to classify between cats, and dogs for my course project in Zero to GANs in Pytorch course.

I choose this topic for my course project because in the past I heard a lot from classifying cats and dogs, and it was somewhat magic to me.

I made my project using the notebooks from the previous lessons. I used the same methods for almost everything but I needed to make some modifications to the code.

First I had to resize the dataset because the picture sizes were inconsistent. To solve this I used Transform.Resize() this took me around 3 hours to find, and then we had almost the same thing in the next lesson.

Next I had to make a train and a validation dataset. The train folder consist of 23000 pictures of cats and dogs. I had 20000 randomly selected pictures for training and 3000 random images for validation. I used the validation folder for testing, it consisted of 2000 random images of cats and dogs.

The model was slightly modified to take in 64x64 images instead of 32x32

The training methods were the same, I set the learning rate to 0.0001 and the number of epochs to 12 because at 13–14 the validation loss got bigger and bigger.

Here are some examples of the test datas.

And here is my notebook.

Related posts:

Some people claim that they lost ten pounds in one week, whereas others lose only one pound each week. We’ll discuss how much weight you can realistically lose on keto over a three month period in…

A slasher satire of art criticism, Velvet Buzzsaw promises bite, blood, and a true element of danger — even if the pretentious victims are less than sympathetic. However, after an explosive trailer…

Selecting the right e-commerce platform is crucial for the success of your online business. Salesforce Commerce Cloud and SAP Hybris Commerce Cloud are two leading contenders in the market. In this…