Pet Pawpularity

Predict pet cuteness scores using machine learning.

Table of Contents

1. About The Project
   • Built With
2. Detailed information
3. Contact

About The Project

Hi, I am Bex! I built this project to create a simple web app that would allow any user to upload an image of their pet and get a cuteness score. The data comes from Petfinder.my website ( source) and contains about 10k images with labels for their cuteness. As cuteness is such a subjective concept, the scores returned from the app are not necessarily accurate. In fact, even the best solutions to this challenge on Kaggle are very close to a solution that returns just random scores.

(back to top)

Built With

• DVC
• MLFlow
• Streamlit
• BentoML
• DagsHub
• Heroku
• TensorFlow

(back to top)

Detailed description of the project

I explain my approach to solve the project in three articles on my Medium blog:

• Part 1: Project Overview and DVC Setup
• Part 2: Detailed tutorial to MLFlow and experiment tracking for the project
• Part 3: In-depth Tutorial to deploying the project with the combination of DagsHub, BentoML, Streamlit

You can also try out the API for this project by sending a POST request to this address. Please, read the last part of the article for the details.

(back to top)

Contact

Bex Tuychiev - @BexTuychiev - bex@ibexprogramming.com

Project Link: https://github.com/BexTuychiev/pet_pawpularity

(back to top)