COMPARATIVE ANALYSIS OF THE ACCURACY AND EFFICIENCY OF MOTION DETECTION TOOLS AND SYSTEMS FOR PIR SENSOR, OPENCV WEBCAM, AND RASPBERRY PI

• Published in: Електроніка та інформаційні технологіі Vol. 30; no. 30; pp. 89 - 98
• Main Authors: Diachok, Roman, Klym, Halyna, Tepliakov, Ivan
• Format: Journal Article
• Language: English
• Published: Ivan Franko National University of Lviv 01.06.2025
• Subjects: computer vision; facial recognition system; opencv; raspberry pi
• ISSN: 2224-087X; 2224-0888; 2224-0888
• DOI: 10.30970/eli.30.7

Background. This paper focuses on developing and evaluating a facial recognition system optimized for real-world conditions. A prototype system was implemented, featuring a face detection algorithm, hardware configuration, and integration of OpenCV, Dlib, and Picamera2 libraries, along with pre-trained models for accurate facial landmark detection. Experimental tests were conducted under various conditions, including lighting changes, different angles, and partial occlusions. Materials and Methods. The study aimed to analyze and select an algorithm for face recognition, considering hardware limitations, ensuring high image processing accuracy and speed, and integrating the proposed solution into the compact and energy-efficient Raspberry Pi platform. The subject of the study involved the development of an efficient and energy-saving system for real-time face detection and recognition, ensuring accuracy, reliability, and high performance under constrained computational resources. Results and Discussion. The study presented the main stages of developing a face recognition system and assessed its performance and resilience under real-world operating conditions. This approach substantially reduced processing time and accelerated the identification procedure during subsequent queries, which is crucial for resource-limited platforms like the Raspberry Pi. Additionally, methods for improving system efficiency were explored through algorithmic optimizations and fine-tuning. The results demonstrated the proposed system's high accuracy and operational stability under favorable conditions, such as frontal face orientation relative to the camera, minimal external interference, and a fixed facial position. Conclusion. The study successfully developed a face recognition system optimized for the Raspberry Pi platform, achieving high accuracy and efficiency despite hardware limitations. Integrating a pre-processed feature descriptor database and algorithmic optimization strategies was key in improving system performance. The proposed solution showed strong potential for real-world deployment, particularly in applications where energy efficiency and compactness are critical.