Treffer: Cost-effective object identification and distance estimation using YOLOv3 on Raspberry Pi: A practical approach.

Distance estimation is critical in various applications such as autonomous vehicles and robotics. This paper presents a cost-effective, portable vision system for real-time object detection and distance estimation, leveraging the YOLOv3 deep learning model on a Raspberry Pi platform. The system integrates object identification and distance measurement using optimized image processing techniques specifically designed for resource-constrained hardware. YOLOv3 was selected for its balance between speed, accuracy, and computational efficiency, making it ideal for applications demanding real-time processing on low-power devices. The Raspberry Pi 3 Model B+ and Pi Cam v1.3 were selected as the hardware platform, with a total system cost of approximately RM289 (USD 62). The system accurately identifies objects within YOLOv3's database, achieving confidence scores of 0.50 or higher, even in challenging scenarios such as objects displayed on digital screens. Distance estimation is performed by calculating the size of the bounding box around the detected object, utilizing a calibrated scaling factor specific to the camera employed. Experimental results demonstrate that the system achieves an error of 7% or less for distances greater than 10 cm, with increased errors at closer distances due to limitations in the camera's field of view. This paper highlights the system's potential for applications in consumer and industrial settings, including personal navigation aids, advanced driver assistance systems, and automated manufacturing. Overall, the study validates the feasibility of deploying YOLOv3 on a Raspberry Pi for cost-effective object detection and distance estimation, providing a practical solution for a range of real-world applications. [ABSTRACT FROM AUTHOR]

Copyright of AIP Conference Proceedings is the property of American Institute of Physics and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)