REDA: A Real-Time Event-Detection Approach To Minimize IoT Visual Data Generation With Computation Efficiency.

Area of Science:

• Computer Science
• Artificial Intelligence
• Internet of Things

Background:

• Excessive visual data from environmental monitoring in IoT systems poses significant challenges.
• Current data minimization methods often compress already generated data, leading to high computational costs and distortion.

Purpose of the Study:

• To introduce REDA, a novel real-time event-driven approach for minimizing visual data generation in IoT.
• To address the limitations of existing methods in terms of real-time reduction, computational overhead, and data distortion.

Main Methods:

• REDA employs an event estimation method integrating motion and multi-scale object detection.
• Utilizes an Optimal-IoU loss function to manage gradient challenges.
• Applies contextual optical flow and filtering techniques for data loss and distortion minimization.

Main Results:

• REDA effectively reduces false alarms and missed detections, lowering computational costs.
• Demonstrates superior real-time data minimization compared to state-of-the-art solutions.
• Achieves high efficiency in visual data reduction.

Conclusions:

• REDA offers an efficient and effective solution for real-time visual data minimization in IoT environmental monitoring.
• The proposed method overcomes the drawbacks of traditional data compression techniques.
• Enables enhanced quality of life through optimized IoT data management.