CIFAR-10 Image Classification Using PyTorch

This project implements a convolutional neural network (CNN) for classifying images from the CIFAR-10 dataset using PyTorch.

Dataset

The model is trained on the CIFAR-10 dataset, which consists of 60,000 32x32 color images in 10 different classes:

• Plane
• Car
• Bird
• Cat
• Deer
• Dog
• Frog
• Horse
• Ship
• Truck

Dependencies

Ensure you have the following libraries installed before running the code:

pip install numpy pillow torch torchvision

Model Architecture

The CNN consists of:

• Two convolutional layers with ReLU activation and max pooling.
• Three fully connected layers to process image features.
• Cross-entropy loss for classification.
• Stochastic Gradient Descent (SGD) optimizer.

Training the Model

Run the following notebook to train the model:

main.ipynb

The model is trained for 30 epochs, and the training loss is printed after each epoch.

Saving & Loading the Model

After training, the model's weights are saved to net.pth: torch.save(net.state_dict(), 'net.pth')

To load the trained model: net.load_state_dict(torch.load('net.pth'))

Evaluating Accuracy

After loading the model, run the evaluation script to compute accuracy on the test dataset: which gave an accuracy of 67.1%

Predicting Images

To classify new images:

1. Place test images inside the test/ directory.
2. Run the test cell on the notebook

License

This project is licensed under the MIT License. Feel free to use and modify as needed.

---

Happy coding! 🚀