This is a simple app to classify dogs using fastai and streamlit. The app is deployed using Streamlit Sharing. Click here to view and play with the app. This project is inspired by Chapter 2 of the book Deep Learning for Coders with fastai & PyTorch.

Find the tutorial on how to create your own dog classifier in this Medium article.

450 different dog images are obtained using Bing Image Search API. There are 150 images of each type of dog. The non-relevant images are removed.

The model was trained to recognize 3 types of dog: Winner, Chihuahua, and Basset Hound using fastai. You can find the details of the training in train_dog_classifier.ipnb notebook. The model was saved to dog.pkl.

Click Browse files to upload a dog image. Note that since the app is trained on just 3 kinds of dogs: Winner, Chihuahua, and Basset, make sure to upload only images of these dogs.

I use my dog's photos and other photos I found from Facebook to test the performance of the model. Note that these images were not used to train the model. The model can recognize dogs with high accuracy. Here are some results I got: