ETIENNE JULIEN - PORTFOLIO

This personal project is aimed at developing a functional generative artificial intelligence using Python and C++. The main goal is to artificially generate images of dogs or cats based on the user's input. By working on this project, my goal is to deepen my understanding of generative models, particularly in the context of image generation, and to explore the intersection of machine learning and computer vision. I have also set up a GitHub repository to make my work accessible and to document my progress : https://github.com/Crodaliox/GenerativeAIProject

In my journey to explore deep learning, I began by developing a Convolutional Neural Network (CNN) in python for image classification using the popular MNIST dataset. MNIST, consisting of handwritten digits, is a benchmark dataset in machine learning and computer vision. The goal was to build a robust model capable of accurately classifying the digits from 0 to 9. Through the implementation of the CNN, I learned the intricacies of feature extraction using convolutional layers, pooling operations, and fully connected layers to predict image classes.

After successfully implementing the image classifier, I advanced my exploration by delving into Generative Adversarial Networks (GANs). Unlike the classifier, which was trained to distinguish between different digits, the GAN model focuses on generating realistic handwritten digits that replicate the MNIST data distribution. A key challenge in this phase of the project was developing a conditional image generation algorithm, which necessitates careful management of labels to produce images based on specific input conditions. I have now succeeded in coding the conditional GAN, and while there are still refinements to be made in achieving precise image generation, I am pleased with the progress I have made.