Fashion MNIST Image Recognition Project with TensorFlow

Description

This project aims to apply and compare various neural network algorithms for images recognition on the Fashion MNIST dataset. It's a project to improve my skills with Tensorflow and Python. Utilizing TensorFlow, a machine learning and neural networks library, this project strives to accurately identify different types of fashion items in the Fashion MNIST dataset.

Requirement

You need Python 3.9 (developped with Python 3.9.13) and pip 24.0.

Installation

Clone this repository to your local machine using:

git clone https://github.com/GuillaumeBrgnFR/FashionMnist.git

Install the necessary dependencies by running:

pip install -r requirements.txt

Create a data directory. Download the dataset at https://www.kaggle.com/datasets/zalando-research/fashionmnist and unzip the file in the data directory.

Project Structure

• data: Directory with the datasets.
• fashion_mnist.ipynb: Run all cells to execute the program with a simple Convolutional Neural Network.
• fashion_mnist.ipynb: Run all cells to execute the program with a tree Convolutional Neural Networks and two different vote systems.
• results: Old executed notebooks in HTML format.

Results

Describe the results obtained with your model here, including:

• Classification accuracy on the test set.
• A discussion on the model's performance.
• Visualizations of the model's predictions (optional).

Conclusion

Summarize key learnings from this project, as well as possible future improvements.

Authors

• Guillaume Bouregon