CIFAR-10 Image Classification using Convolutional Neural Network (CNN)

Overview

This project aims to build a Convolutional Neural Network (CNN) for image classification using the CIFAR-10 dataset. The CIFAR-10 dataset consists of 60,000 color images of size 32 x 32 distributed over 10 classes, making it a widely used benchmark in research for image classification tasks.

Steps to Reproduce

1. Load and Normalize CIFAR-10 Dataset

Download the CIFAR-10 dataset from the official website: CIFAR-10 Dataset. The dataset is split into 50,000 training images and 10,000 test images. Normalize the pixel values of the images.

2. Define CNN Model, Loss Function, and Optimizer

Define a Convolutional Neural Network (CNN) model suitable for image classification. Choose an appropriate loss function and optimizer for training the model. Consider using techniques such as dropout to reduce overfitting.

3. Data Augmentation

Implement data augmentation using torchvision.transforms to artificially increase the diversity of the training dataset. This helps in improving the generalization ability of the model.

4. Training the Model

Train the CNN model on the training dataset. Monitor training and validation loss and accuracy during the training process.

5. Evaluation

Test the trained model on the test dataset and report the test accuracy. Evaluate the model's performance on unseen data.

6. Plotting Model Curves

Plot the training and validation curves for both loss and accuracy. Visualize how these metrics change over epochs to analyze the model's learning progress.

Requirements

• Python (>=3.6)
• PyTorch (>=1.0)
• torchvision
• matplotlib

License

This project is licensed under the MIT License.