Data_Classification_Preparation_Training

Complete dataset cleaning, augmentation, and preprocessing pipeline

This repository provides a complete image dataset preparation pipeline for machine learning classification tasks.
It is divided into three major modules:

1. C1_Dataset_Cleaning – Cleans and validates image data.
2. C2_Data_Augmentation – Balances classes using data augmentation.
3. C3_Splitting_Preprocessing – Splits and preprocesses the dataset for training.

---

Features

• Detects and removes corrupted or invalid image files.
• Converts image formats to a consistent standard (.jpg).
• Balances dataset classes using augmentation.
• Splits dataset into train, validation, and test sets.
• Resizes and prepares images for model input.

---

Folder Structure

---

Prerequisites

• Python 3.10 or higher
• PIL (Pillow), tqdm, json, os libraries
• Google Colab or local Python environment
• Required libraries: os, cv2, numpy, PIL, shutil, json, random

---

Setup Instructions

# Clone this repository
git clone https://github.com/sreyasudevan/Data_Classification_Preparation_Training.git
cd Data_Classification_Preparation_Training

# Install dependencies
pip install -r requirements.txt