Capstone project for the Machine Learning Nanodegree
The leaf classification project is based on a Kaggle competition where participants are challenged to produce a machine learning model capable of classifying 99 different plant species based on leaf images and features extracted from these images . The challenge is to have the minimum classification error possible. The ultimate solution should be able to classify all leaves correctly. The classification challenge can be improved in future extensions by including more plant species.
In this project, I plan to solve this challenge by first loading the image data and the extracted features, followed by data exploration and preprocessing, and finally applying different machine learning algorithms such as decision trees, Adaboost, Support Vector Machines, K-Nearest Neighbors, and Convolutional Neural Networks. I hope that one or a combination of these powerful algorithms will be able to successfully solve this challenge by accurately predicting most or all of the provided test images, which is achieved by minimizing the log loss metric.
This project uses the following python packages:
Python 3.6 sklearn version 0.19.1 keras version 2.0.9 matplotlib version 2.1.0 numpy version 1.12.1 pandas version 0.20.3 seaborn version 0.8.1 zipfile version unknown - from Udacity workspace