CIFAR-10 CNN Image Classification

This project implements a Convolutional Neural Network (CNN) to classify images from the CIFAR-10 dataset using PyTorch.

Project Structure

• data_loader.py: Handles dataset downloading, augmentation, and loading.
• model.py: Defines the CNN architecture (4 convolutional layers + 2 fully connected layers).
• train.py: Script to train the model, saves the best weights to best_model.pth.
• evaluate.py: Generates training loss/accuracy plots and visualizes sample predictions.
• requirements.txt: Project dependencies.

Setup and Usage

1. Install Dependencies:

   pip install -r requirements.txt
   

2. Train the Model:

   python train.py --epochs 10
   

3. Evaluate and Visualize:

   python evaluate.py
   

Deployment

1. Install Dependencies:

   pip install -r requirements.txt
   

2. Run in Development Mode:

   python app.py
   

3. Run in Production Mode (Windows):

   python app.py --prod
   

The web interface is available at http://127.0.0.1:5000. It features:

• Drag and Drop: Easily upload images for classification.
• Real-time Prediction: Deep learning inference results in seconds.
• Modern UI: Glassmorphism design with responsive elements.

Results

After training, you can find:

• training_curves.png: Plots showing accuracy and loss over epochs.
• sample_predictions.png: A grid of test images with predicted vs. true labels.