CIFAR-10 Image Classification Tutorial

This repository demonstrates deep learning techniques for image classification using the CIFAR-10 benchmark dataset. Through a series of progressively more sophisticated models, it shows how different neural network architectures and training techniques improve classification performance.

Package: The tools developed for this project are available as image-classification-tools on PyPI for use in your own projects.

Documentation: gperdrizet.github.io/CIFAR10

About the dataset

CIFAR-10 consists of 60,000 32×32 color images across 10 classes:

• Training set: 50,000 images (40,000 for training, 10,000 for validation)
• Test set: 10,000 images
• Classes: airplane, automobile, bird, cat, deer, dog, frog, horse, ship, truck

The models

1. 01-DNN.ipynb - Baseline fully-connected deep neural network on grayscale images
2. 02-CNN.ipynb - Convolutional neural network to exploit spatial relationships
3. 03-RGB-CNN.ipynb - CNN modified to process full RGB color information
4. 04-optimized-CNN.ipynb - Bayesian hyperparameter optimization with Optuna
5. 05-augmented-CNN.ipynb - Data augmentation for improved generalization

Getting started

Prerequisites

• Docker
• Visual Studio Code
• Dev Containers extension

Setup

1. Clone the repository:

git clone https://github.com/gperdrizet/CIFAR10.git
cd CIFAR10

2. Open in VS Code:

code .

3. When prompted, click "Reopen in Container" or use the command palette (Ctrl+Shift+P / Cmd+Shift+P) and select "Dev Containers: Reopen in Container"

4. Wait for the container to build and initialize (first time only). The devcontainer will:

   • Set up the Python environment
   • Install all dependencies
   • Download the CIFAR-10 dataset

Running the notebooks

Once the container environment loads, you should be all set - you can start running the notebooks! Note: run them in order - some depend on output generated by earlier notebooks.

Pre-trained models

Models from each experiment are saved in models/pytorch/:

• dnn.pth - Baseline DNN
• cnn.pth - Basic CNN
• rgb_cnn.pth - RGB CNN
• optimized_cnn.pth - Hyperparameter-optimized CNN
• augmented_cnn.pth - Optimized CNN trained on augmented data

Environment

All experiments use:

• Python 3.10+
• PyTorch 2.0+
• NVIDIA GPU with CUDA support (when available)

Dependencies

Core requirements:

• python ≥3.10, <3.13
• torch ≥2.0
• torchvision ≥0.15
• numpy ≥1.24
• matplotlib ≥3.7
• optuna ≥3.0
• scikit-learn ≥1.3

See pyproject.toml for complete dependency specifications.

Citation

If you find this project helpful for your learning or work, feel free to reference it:

@software{cifar10_tutorial,
  author = {Perdrizet, George},
  title = {CIFAR-10 Image Classification Tutorial},
  year = {2024},
  url = {https://github.com/gperdrizet/CIFAR10}
}

License

This project is licensed under the GPLv3 License - see the LICENSE file for details.