CateOrDoge

Code for a website that contains image classifier. Deployed using Flask on Heroku, styled with Bootstrap

Thank you for taking interest in this website!

This website was created with Flask, hosted on Heroku and styled with Bootstrap. Classifies the image you upload, and tells you whether the image you have uploaded is a cat or a dog, amongst other things.

1. Step 1: Upload your image (png / jpg / jpeg) to the website
2. Step 2: Use either ResNetV250 or MobileNetV2 (with transfer learning) to predict your image
3. Step 3: The website will update the background image, and return what it thinks the iamge contains (In the case of ResNetV2: tabby, humming bird etc)

Additional Comments

I used transfer learning on MobileNetV2, adding additional layers to the model. While the main model remained untrained, I trained the added layers on cifar10. Hence, the MobileNetV2 option return 10 classes which are airplanes, cars, birds, cats, deer, dogs, frogs, horses, ships, and trucks.

You can check out how I trained the model here:
https://drive.google.com/file/d/10z9Ia27-rguGbsmwSRDnXnFf-aAyF6ex/view?usp=sharing