✅ CNN_Pytorch_Demo

CNN Image Classification with PyTorch

This repository demonstrates a simple convolutional neural network (CNN) implemented in PyTorch for image classification. It covers model definition, training, evaluation, and visualization of results.

Project Overview

• Builds a custom CNN model using PyTorch
• Trains on an image dataset using DataLoader and torchvision.transforms
• Evaluates accuracy on test data
• Visualizes predictions and performance metrics

Tech Stack

• Python
• PyTorch
• torchvision
• matplotlib, numpy

How to Run

1. Clone the repository: git clone https://github.com/Ghostdevc/CNN_Pytorch_Demo.git cd CNN_Pytorch_Demo

2. Install dependencies: pip install -r requirements.txt

3. Prepare your image dataset:

• Modify the dataset loading path inside main.py
• Expected folder structure:

  data/
    train/
      class1/
      class2/
    test/
      class1/
      class2/
  

4. Run the training script: python main.py

Main Components

• Custom CNN model subclassing nn.Module
• Data augmentation and normalization using torchvision.transforms
• Training loop with loss/accuracy tracking
• Confusion matrix and prediction visualization

Notes

• Easily adaptable for CIFAR-10, MNIST, or custom image datasets
• Great for understanding the structure of a basic CNN in PyTorch

License

MIT License