Flask-Image-Classifier

A simple Flask web app for real-time image classification using a pretrained MobileNetV2 deep learning model (ImageNet weights).

Users can upload an image through the web interface, and the app predicts the most likely class with confidence.

---

Features

• Upload any image via the web interface
• Real-time classification using MobileNetV2
• Displays predicted label with confidence percentage
• Web UI styled with Bootstrap
• Easy to run locally

---

Project Structure

Flask-Image-Classifier/
├── app.py # Flask application
├── templates/
│ └── index.html # Frontend (Bootstrap upload form)
└── images/ # Uploaded images (auto-saved here)

---

Requirements

• Python 3.8+
• Install dependencies:

  pip install flask tensorflow
  

---

Usage

1. Clone the repo

git clone https://github.com/Anjanamb/Flask-Image-Classifier.git
cd Flask-Image-Classifier

2. Run the Flask app

python app.py

3. Open in browser

http://127.0.0.1:3000/

4. Upload an image

• Select an image file.
• Click Predict Image.
• The prediction and confidence will be displayed.

---

Model Details

• Uses MobileNetV2 pretrained on ImageNet (1000 classes).
• Image is resized to 224x224, preprocessed, and fed to the model.
• Output is decoded using decode_predictions.

Example prediction output:

Labrador retriever (95.23%)

---

Future Improvements

• Support for multiple top predictions
• Display uploaded image alongside prediction
• Add Docker support for easier deployment
• Deploy on cloud (Heroku, AWS, etc.)
• Option to fine-tune MobileNetV2 with custom dataset

---

License

This project is licensed under the MIT License. See the LICENSE file for details.

---