Abstract
Realtime image detection and recognition on low-power edge devices are key to implementing many social applications such as automated traffic guard systems. This paper proposes mobile-oriented convolutional neural network implementation for object detection, focusing on computational reduction techniques such as depthwise separable convolution and grouped convolution on an edge GPU device. Through empirical evaluation and analysis, it is shown that the use of grouped convolution and half-precision floating point arithmetic is effective for the calculation of \(3 \times 3\) convolution, and the grouped convolution with 8 groups for YOLOv3-tiny achieves 16 FPS with 86% detection accuracy in a 5 W low power mode on Jetson Nano. We also discuss how the processing time and recognition accuracy are affected by the floating arithmetic types and power consumption modes that the GPU device offers.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Bochkovskiy, A., Wang, C., Liao, H.M.: YOLOv4: optimal speed and accuracy of object detection. arXiv preprint arXiv:2004.10934 (2020)
Feng, H., Mu, G., Zhong, S., Zhang, P., Yuan, T.: Benchmark analysis of YOLO performance on edge intelligence devices. Cryptography 6(2), 1–16 (2022)
Google: Open images dataset. https://storage.googleapis.com/openimages/web/index.html. Accessed 30 Mar 2023
Howard, A.G., et al.: MobileNets: efficient convolutional neural networks for mobile vision applications. arXiv preprint arXiv:1704.04861 (2017)
Jiao, L., et al.: A survey of deep learning-based object detection. IEEE Access 7, 128837–128868 (2019)
Kingma, D.P., Ba, J.: Adam: a method for stochastic optimization. In: Proceedings of the International Conference on Learning Representations (ICLR), pp. 1–15 (2015)
Krizhevsky, A., Sutskever, I., Hinton, G.E.: ImageNet classification with deep convolutional neural networks. In: Proceedings International Conference on Neural Information Processing Systems, pp. 1097–1105 (2012)
Redmon, J.: YOLO: real-time object detection. https://pjreddie.com/darknet/yolo/. Accessed 30 Mar 2023
Redmon, J., Divvala, S.K., Girshick, R.B., Farhadi, A.: You only look once: unified, real-time object detection. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 779–788 (2016)
Redmon, J., Farhadi, A.: YOLOv3: an incremental improvement. arXiv preprint arXiv:1804.02767 (2018)
Tajar, A.T., Ramazani, A., Mansoorizadeh, M.: A lightweight tiny-YOLOv3 vehicle detection approach. J. Real-Time Image Proc. 18, 2389–2401 (2021)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Araki, Y., Kawazu, T., Manabe, T., Ishizuka, Y., Shibata, Y. (2023). A Mobile-Oriented GPU Implementation of a Convolutional Neural Network for Object Detection. In: Barolli, L. (eds) Complex, Intelligent and Software Intensive Systems. CISIS 2023. Lecture Notes on Data Engineering and Communications Technologies, vol 176. Springer, Cham. https://doi.org/10.1007/978-3-031-35734-3_15
Download citation
DOI: https://doi.org/10.1007/978-3-031-35734-3_15
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-35733-6
Online ISBN: 978-3-031-35734-3
eBook Packages: Intelligent Technologies and RoboticsIntelligent Technologies and Robotics (R0)