Potato-tomato classifier

It's a basic classifier trained with ResNet18 on pictures obtained using Bing Search API. It was inspired by and heavily based on fast.ai lessons (available here: https://course.fast.ai) and the book "Deep Learning for Coders with Fastai and PyTorch: AI Applications Without a PhD", which you can buy or read as interactive Jupyter Notebooks for free.

In order to "play" with the potato-tomato classifier, you can:

• use the notebook (you will have to download the export.pkl file and then uncomment the commented lines and run it all)
• use this link to launch the notebook as an app (be patient, it takes a while). Once it's ready, it should look like this:
• if the link above fails, you can manually deploy and use the notebook as an app (for free) using Binder
  1. go to https://mybinder.org
  2. write GJuceviciute/potatotomato-classifier under GitHub repository name or URL
  3. choose URL under Path to a notebook file (it will change from Path to a notebook file to URL to open)
  4. write /voila/render/Potato_tomato_classifier.ipynb under the URL to open
  5. Press launch and wait
  6. Follow the instructions there, i.e., upload a picture, and wait for the magic - prediction results