♻️ Waste Product Classification

Transfer Learning & Fine-Tuning | Flask Deployment

📌 Project Summary (ATS-Optimized)

An end-to-end computer vision and deep learning project that classifies waste images into multiple categories using transfer learning and fine-tuning of a pre-trained CNN (VGG16).
The trained model is deployed using a Flask web application that supports real-time image upload, preprocessing, prediction, and confidence scoring, enabling automation in waste management and sustainability systems.

🚀 Key Skills Demonstrated (ATS Keywords)

• Computer Vision
• Deep Learning
• Transfer Learning
• Convolutional Neural Networks (CNN)
• TensorFlow / Keras
• Flask Deployment
• Image Classification
• Model Fine-Tuning
• Data Augmentation
• Model Inference Pipeline
• Sustainable AI Applications

🧠 Project Features

• Transfer learning using VGG16 pretrained on ImageNet
• CNN fine-tuning for domain-specific waste classification
• Automated dataset splitting (train / validation / test)
• Image preprocessing pipeline (resize, normalization, batching)
• Real-time inference via Flask web application
• Prediction confidence score output
• Modular, production-ready project structure

🗂️ Waste Categories

The model classifies waste images into the following categories:

• Cardboard
• Glass
• Metal
• Paper
• Plastic
• Trash

🛠️ Technology Stack

📁 Project Structure

waste_product_classification/
│
├── app.py
├── requirements.txt
├── README.md
│
├── models/
│   ├── model.h5
│   └── labels.txt
│
├── data/
│   ├── train/
│   ├── val/
│   └── test/
│
├── raw_dataset/
│
├── static/
│   ├── css/
│   │   └── style.css
│   └── uploads/
│
├── templates/
│   └── index.html
│
├── utils/
│   ├── preprocess.py
│   └── model_loader.py
│
├── notebooks/
│   └── training.ipynb
│
└── split_dataset.py

⚙️ Workflow Overview

1. User uploads a waste image through the web interface
2. Image preprocessing (resizing, normalization, batching)
3. CNN model performs inference
4. Predicted waste class and confidence score are displayed

📊 Model Training Details

• Pre-trained VGG16 model initialized with ImageNet weights
• Custom dense classification head added
• Data augmentation to improve generalization
• Fine-tuning of upper convolutional layers
• Adam optimizer with categorical cross-entropy loss

📸 Application Screenshots

Home Page & Upload Interface

Image Preview

Prediction Output

▶️ How to Run

git clone <your-repo-url>
cd waste_product_classification
python -m venv venv
venv\Scripts\activate
pip install -r requirements.txt
python app.py

Access the application at:

http://127.0.0.1:5000/

🌍 Real-World Use Cases

• Smart waste segregation systems
• Recycling and sustainability automation
• AI-powered environmental monitoring
• Intelligent waste management pipelines

📌 Resume-Ready Highlights

• Built an end-to-end waste classification system using deep learning and computer vision
• Applied transfer learning and fine-tuning on real-world waste image datasets
• Deployed a CNN model using Flask for real-time image inference
• Designed a modular and production-ready ML deployment pipeline

📄 License

This project is intended for educational and portfolio purposes.