Automatic Change Detection System over Unmanned Aerial Vehicle Video Sequences Based on Convolutional Neural Networks

Area of Science:

• Computer Vision
• Artificial Intelligence
• Robotics

Background:

• Unmanned aerial vehicles (UAVs) are increasingly used for surveillance, requiring automated image processing for tasks like change detection.
• Existing methods struggle with variations in UAV flight paths and environmental conditions, impacting detection accuracy.

Purpose of the Study:

• To develop and evaluate a robust change detection system for video sequences acquired by moving cameras, specifically addressing UAV surveillance challenges.
• To enhance the adaptability and precision of change detection algorithms in dynamic environments.

Main Methods:

• A system combining image alignment techniques with a deep learning model based on convolutional neural networks (CNNs).
• Testing on the Change Detection 2014 dataset, including challenging scenarios like dynamic backgrounds, intermittent object motion, and adverse weather.

Main Results:

• The system achieved a mean F-measure score exceeding 97% across various scenarios.
• Demonstrated significant precision in detecting changes in intermittent object motion (above 99%) and robust performance in dynamic and adverse weather conditions.

Conclusions:

• The proposed system offers a precise and adaptable solution for change detection in UAV-based surveillance.
• The approach effectively handles variations in UAV flight and environmental factors, outperforming state-of-the-art methods in challenging scenarios.