The inspiration for this project mainly came from wanting to learn Machine Learning image classification.
Nota.dog takes a photo and runs it through a Machine Learning algorithm and then tells you if it's a dog. And because this was made at the Corgi Hacks hackathon it can also tell if it’s a corgi.
The app was made with react.js and is hosted on Google Cloud. I used TensorFlow and Teachable Machine for the Machine Learning algorithm. It was trained using three sets of 500 hand-picked images one for dogs, one for corgis, and one for not a dog. I also used React-Bootstrap for the styling.
The two biggest challenges I ran into were with training the machine and processing the images in the browser. I also had a hard time centering things, but who doesn’t.
I’m proud of how well the algorithm can tell the difference between classes (Dog or not dog). I also am really happy with how the app looks.
The biggest thing I learned was how to train and implement the Machine Learning algorithm. I also learned about how to handle and process images in the browser.
The next thing I would like to implement is the ability to tell different breeds of dogs apart. I would also love to improve the algorithm.