CNN Multi-Class Classifier 🧠🔍

📋 Table of Contents

• Project Overview
• Features
• Project Structure
• Installation
• Usage
• Model Architecture
• Training
• Prediction
• Performance
• Hardware Requirements
• Contribution

🌟 Project Overview

This project implements a Convolutional Neural Network (CNN) for multi-class classification. The model is built from scratch, providing a deep understanding of the underlying architecture and training process. With an accuracy of around 90%, it demonstrates strong performance on the classification task.

🚀 Features

• Custom CNN model implementation
• Defined training loop
• Configurable loss function and optimizer
• Model training and accuracy evaluation
• Model saving functionality
• Separate prediction script for the saved model
• High accuracy (~90%)
• CPU-based training, no GPU required

📁 Project Structure

The project is organized into separate notebooks for better modularity:

• model.ipynb: Contains the CNN model architecture & it includes the training loop, loss function, and optimizer
• predict.ipynb: Handles predictions using the saved model

🛠️ Installation

git clone https://github.com/akhil-varsh/cnn-multi-class-classifier.git
cd cnn-multi-class-classifier
pip install -r requirements.txt

🖥️ Usage

To train the model:

1. Run model.ipynb to define the model architecture

To make custom predictions using the saved model: Run predict.ipynb

🏗️ Model Architecture

The CNN architecture is defined from scratch in model.ipynb. It includes:

• Convolutional layers
• Pooling layers
• Fully connected layers
• Activation functions

For detailed architecture, refer to the model.ipynb file.

🏋️ Training-

The training process is implemented in model.ipynb and includes:

• Custom training loop
• Loss function: CrossEntropyLoss ---> Epoch [1/40], Loss: 1.2354 Epoch [40/40], Loss: 0.0000
• Number of epochs: 40

🔮 Prediction

The predict.ipynb notebook is provided for making predictions using the saved model. This notebook loads the trained model and can be used for inference on new data.

Here's an example of a prediction made by the model:

📈 Performance

The model achieves an accuracy of approximately 90% on the test dataset, demonstrating its effectiveness in multi-class classification tasks.

💻 Hardware Requirements

This project is designed to run on standard CPU hardware. No GPU is required for training or inference, making it accessible to a wide range of users without the need for specialized hardware.

🤝 Contribution

Contributions to improve the project are welcome.

---

Made with ❤️ by the Akhil