Real-time Image Classification with MobileNetV2

This project demonstrates real-time image classification using a webcam feed and the pre-trained MobileNetV2 model from TensorFlow Keras.

📌 Features

• Loads MobileNetV2 pre-trained on ImageNet.
• Captures live video from webcam.
• Classifies each frame in real-time.
• Displays predicted label and confidence score on screen.
• Press 'q' to quit.

🛠️ Requirements

Install dependencies before running the program:

pip install opencv-python tensorflow numpy

▶️ Usage

Run the script:

python RealTimeImageClassification.py

Press 'q' to exit the window.

📂 File Structure

.
├── RealTimeImageClassification.py  # Main script
└── README.md                                # This documentation

⚙️ How It Works

1. Loads MobileNetV2 with pre-trained ImageNet weights.
2. Captures video from the webcam.
3. Preprocesses each frame:
   • Converts BGR → RGB
   • Resizes to 224×224
   • Applies MobileNetV2 preprocessing
4. Passes the frame through the model.
5. Displays the top predicted class with confidence score.

🧠 Model

• MobileNetV2 -- Lightweight convolutional neural network optimized for real-time applications.
• Pre-trained on ImageNet (1,000 classes).

🖼️ Output Example

When you point the webcam at an object, you'll see:

Laptop : 95.2%

overlayed on the live feed.

🔑 Notes

• Ensure your webcam is accessible.
• First prediction might be slightly slower due to model loading.
• Works best in good lighting conditions.

👤 Author

Developed by Prithviraj Chavan.