Real-Time Motion Detection System for Automatic Surveillance

A real-time motion detection system built using Kalman Filtering and Convolutional Neural Networks (CNNs) to enhance the accuracy of motion detection while reducing false positives. The system also employs Gaussian blur to minimize environmental noise, improving the reliability of the detection in low-quality video footage.

Features

• Kalman Filtering: Enhances the accuracy of motion detection by predicting object movement and smoothing the detected positions.
• CNN Integration: Improves the detection of moving objects in the video frames, leveraging deep learning for feature extraction.
• Gaussian Blur: Reduces noise in video input, minimizing false detections in challenging environments.
• Performance Metrics: Assesses detection performance using confusion matrices, precision, F1 score, and Signal-to-Noise Ratio (SNR) to ensure real-world effectiveness.

Tech Stack

• Languages: Python
• Libraries: OpenCV, TensorFlow, Keras, NumPy, SciPy
• Tools: Jupyter Notebook (for testing and visualization)