CIFAR-100 Image Classification with PyTorch

This repository focuses on classifying images from the CIFAR-100 dataset using convolutional neural networks (CNNs) built with the PyTorch library. The CIFAR-100 dataset is a challenging benchmark for image classification, containing 100 diverse classes grouped into 20 superclasses.

Dataset Overview

• Training and Testing Images: Each class contains 500 training images and 100 testing images.
• Classes and Superclasses: The dataset includes 100 fine-grained classes, further grouped into 20 broader superclasses.
• Labeling:
  • Fine labels: Specify the exact class of the image.
  • Coarse labels: Indicate the superclass to which the image belongs.

Project Objectives

1. Dataset Division: The training dataset is divided into two subsets:
   • A sub-training set.
   • A validation set (20% of the training data).
2. Fine Label Prediction: The model is designed to predict the fine labels (class) rather than the coarse labels (superclass).
3. Model Experimentation:
   • Experiment with various activation functions, optimizers, hyperparameters, and architectures.
4. Model Selection:
   • Identify the top three models based on performance on the validation set.
5. Full Training:
   • Retrain the top three models using the entire training dataset.
6. Accuracy Testing:
   • Evaluate and compare test accuracy for each model.
7. Benchmarking:
   • Compare model performance against others on the CIFAR-100 leaderboard (considering only models without additional training data).
8. Reporting:
   • Provide a detailed report covering:
     • Activation functions, optimizers, hyperparameters, and architectures used.
     • Test accuracy results.
     • Benchmarking outcomes.
     • Total number of parameters for each model.