👉 Image Classification for Hand-drawn Images from Pictionary Game

GitHub Repo

Introduction

This project was a Kaggle competition where the aim was to make accurate predictions on hand-drawn images of people playing Pictionary game, using different types of ML models for image classification. 6 types of images were selected as the task: banana, boomerang, cactus, crab, flip flops and kangaroo.

The best performing solution was CNNs

Architecture of CNNs

Report & Code

Project Report

Code for CNNs

Code for Random Forests

Code for XGBoost