Home // SIGNAL 2019, The Fourth International Conference on Advances in Signal, Image and Video Processing // View article
Authors:
Samira Karimi Mansoub
Rahem Abri
Anıl Hakan Yarıcı
Keywords: Real-Time Object Detection; YOLO; Multi-Thread Approaches.
Abstract:
Recently, You Look Only Once, version 3 (YOLOv3) approach has been presented as a more efficient solution in the process of object detection. Despite the fact that YOLOv3 can obtain faster and more accurate results than other approaches, it needs to be used in a system with a single powerful Graphics Processing Unit (GPU). However, sometimes, there is a need to process multiple real-time object detection algorithms concurrently on a single GPU, where each object detection algorithm receives a live stream from a camera. It is challenging to have concurrent object detection from live streams on a single GPU. In this paper, we propose a two-step solution to this problem. In the first step, our goal is to provide a model to optimize memory usage and, in the second step, we propose a multi-thread approach that uses YOLOv3 to perform real-time object detection on multiple, concurrent, live streams on a single GPU. In this approach, GPU resources are optimally used. The proposed approach is evaluated on a public dataset and the result shows improvements in performance by an average of 12% in Central Processing Unit (CPU) usage and 13% in frames per second (FPS) compared to the YOLOv3.
Pages: 23 to 28
Copyright: Copyright (c) IARIA, 2019
Publication date: June 2, 2019
Published in: conference
ISSN: 2519-8432
ISBN: 978-1-61208-716-0
Location: Athens, Greece
Dates: from June 2, 2019 to June 6, 2019