Problem type

Supervised learning classification

Models

Neural networks (CNN)

Project realisation date

July 2019

This project is the second of a series of seven projects to be delivered as part of the Udacity DataScience Nanodegree self-study.

It is a supervised learning classification problem in which we have to develop an image classifier with PyTorch. For that we use Tranfer Learning technique: we build a new classifier by using already existing models and replacing the last classifier layer by a new one we train for our specific task. This model has been trained on GPU and is able to classify flowers images among 102 possible classes (the flowers chosen to be flower commonly occuring in the United Kingdom).

The output is something like:

The project contains 2 python scripts:

  • one to train the model
  • one to used a previously trained model to perform prediction

For more details, please refer to my Github repository for this project.

PS: Before enrolling myself in the Udacity nanodegree, I previously made an image classifier with Keras in order to be able to classify dogs vs. cats: here.