Image Classification Experiments

This repository contains Python scripts for experimenting with image classification using both deep learning (ResNet) and traditional machine learning (PCA + k-NN) techniques. The code includes implementations for training a ResNet model on a custom dataset, applying PCA for dimensionality reduction followed by k-NN classification, and visualizing model performance with tools like Grad-CAM.

The dataset used in these scripts is assumed to be an image folder structure (e.g., ./data/train, ./data/valid, ./data/test) with subdirectories for each class, such as a butterfly species dataset. Adjust the paths and parameters as needed for your specific use case.

Table of Contents

• Features
• Requirements
• Installation
• Dataset
• Usage
  • Script 1: ResNet Classification
  • Script 2: PCA + k-NN Classification
  • Script 3: Enhanced ResNet with Grad-CAM
• Contributing
• License

Features

• Deep Learning: Trains a ResNet model with residual blocks for image classification.
• Traditional ML: Implements PCA for dimensionality reduction and k-NN for classification.
• Visualization: Includes loss/accuracy plots, confusion matrices, and Grad-CAM heatmaps for interpretability.
• Data Preprocessing: Applies image transformations like resizing, normalization, and augmentation.

Requirements

• Python 3.8+
• Libraries:
  • torch (PyTorch)
  • torchvision
  • numpy
  • matplotlib
  • scikit-learn
  • tqdm
  • Pillow (PIL)
  • opencv-python (cv2)
  • torchcam (for Grad-CAM visualization)

Installation

1. Clone this repository:

   git clone https://github.com/your-username/your-repo-name.git
   cd your-repo-name