Resource Efficient AI Service Framework Associated with a Real-Time Object Detector
1
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
This paper deals with a resource efficient artificial intelligence (AI) service architecture for multi-channel videostreams. As an AI service, we consider the object detection model, which is the most representative for videoapplications. Since most object detection models are basically designed for a single channel video stream, theutilization of the additional resource for multi-channel video stream processing is inevitable. Therefore, wepropose a resource efficient AI service framework, which can be associated with various AI service models. Our framework is designed based on the modular architecture, which consists of adaptive frame control (AFC)Manager, multiplexer (MUX), adaptive channel selector (ACS), and YOLO interface units. In order to run onlya single YOLO process without regard to the number of channels, we propose a novel approach efficientlydealing with multi-channel input streams. Through the experiment, it is shown that the framework is capableof performing object detection service with minimum resource utilization even in the circumstance of multichannel streams. In addition, each service can be guaranteed within a deadline.
Abstract
This paper deals with a resource efficient artificial intelligence (AI) service architecture for multi-channel videostreams. As an AI service, we consider the object detection model, which is the most representative for videoapplications. Since most object detection models are basically designed for a single channel video stream, theutilization of the additional resource for multi-channel video stream processing is inevitable. Therefore, wepropose a resource efficient AI service framework, which can be associated with various AI service models. Our framework is designed based on the modular architecture, which consists of adaptive frame control (AFC)Manager, multiplexer (MUX), adaptive channel selector (ACS), and YOLO interface units. In order to run onlya single YOLO process without regard to the number of channels, we propose a novel approach efficientlydealing with multi-channel input streams. Through the experiment, it is shown that the framework is capableof performing object detection service with minimum resource utilization even in the circumstance of multichannel streams. In addition, each service can be guaranteed within a deadline.
참고문헌 (10)
- "Yolov4: optimal speed and accuracy of object detection" (2020)[URL]
- Jyoti Parsola, "Post Event Investigation of Multi-stream Video Data Utilizing Hadoop Cluster", International Journal of Electrical and Computer Engineering (IJECE), 8(6), pp.5089~5097 (2018), Institute of Advanced Engineering and Science[DOI]
- S. Karimi-Mansoub, "Concurrent real-time object detection on multiple live streams using optimization CPU and GPU resources in YOLOv3", Proceedings of the 4th International Conference on Advances in Signal, pp.23~28 (2019)
- L. Tan, "A multiple object tracking algorithm based on YOLO detection", Proceedings of 2018 11th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics (CISP-BMEI), pp.1~5 (2018)
- D. Kim, "An intelligent collaboration framework between edge camera and video analysis system", Proceedings of 2018 International Conference on Electronics, Information, and Communication (ICEIC), pp.1~3 (2018)
- G. Mattela, "Enterprise class deep neural network architecture for recognizing objects and faces for surveillance systems", Proceedings of 2019 11th International Conference on Communication Systems & Networks (COMSNETS), pp.607~612 (2019)
- S. Aslan, "Performance analysis of ARM big.LITTLE architecture based mobile processor with multi-thread face detection", Proceedings of 2019 4th International Conference on Computer Science and Engineering (UBMK), pp.336~339 (2019)
- X. Wang, "An efficient end-to-end object detection pipeline on GPU using CUDA", master, pp.2019
- Shuai Liu, "A Robust Parallel Object Tracking Method for Illumination Variations", Mobile Networks and Applications, 24(1), pp.5~17 (2018), Springer Science and Business Media LLC[DOI]
- Jeonghun Lee, "YOLO with adaptive frame control for real-time object detection applications", Multimedia Tools and Applications, 81(25), pp.36375~36396 (2021), Springer Science and Business Media LLC[DOI]