Fashion MNIST Image Classifier

This project implements a multi-class image classifier for the Fashion MNIST dataset using a Convolutional Neural Network (CNN) and Streamlit for the web interface. Users can upload images to classify them into one of ten fashion categories.

Table of Contents

• Requirements
• Installation
• Usage
• Model Architecture
• Image Preprocessing
• License

Requirements

To run this project, you'll need the following Python libraries:

• streamlit
• torch
• torchvision
• torchmetrics
• Pillow
• numpy

You can install the required packages using pip:

pip install streamlit torch torchvision torchmetrics Pillow numpy

Installation

1. Clone this repository to your local machine:

   git clone <repository-url>

2. Navigate to the project directory:

   cd <project-directory>

3. Install the required libraries as mentioned above.

Usage

1. Run the Streamlit app:

   streamlit run app.py

2. Open your web browser and go to the URL provided by Streamlit (usually http://localhost:8501).

3. Use the file uploader to upload an image of clothing. The model will predict the class of the uploaded image.

Model Architecture

The CNN architecture consists of the following layers:

• Convolutional Layer: Applies 16 filters of size 3x3 with ReLU activation and padding.
• Max Pooling Layer: Reduces dimensionality with a kernel size of 2x2.
• Fully Connected Layer: Outputs class predictions for the 10 categories in Fashion MNIST.

Image Preprocessing

Uploaded images are preprocessed using the following steps:

1. Convert the image to grayscale.
2. Resize the image to 28x28 pixels.
3. Normalize the pixel values to be between -1 and 1.

License

This project is licensed under the MIT License - see the LICENSE file for details.

---