2024, Volume 1, Issue 1: 1-16. DOI: 10.62762/TETAI.2024.89422

Research Article | Feature Paper | 07 April 2024
1 College of Architecture and Design, Tongmyong University, Busan 608-711, Korea
* Corresponding Author
Received: 17 December 2023, Accepted: 02 April 2024, Published: 07 April 2024

Abstract
With the rapid development of autonomous driving technology, the demand for real-time and efficient object detection systems has been increasing to ensure vehicles can accurately perceive and respond to the surrounding environment. Traditional object detection models often suffer from issues such as large parameter sizes and high computational resource consumption, limiting their applicability on edge devices. To address this issue, we propose a lightweight object detection model called YOLOv8-Lite, based on the YOLOv8 framework, and improved through various enhancements including the adoption of the FastDet structure, TFPN pyramid structure, and CBAM attention mechanism. These improvements effectively enhance the performance and efficiency of the model. Experimental results demonstrate significant performance improvements of our model on the NEXET and KITTI datasets. Compared to traditional methods, our model exhibits higher accuracy and robustness in object detection tasks, better addressing the challenges in fields such as autonomous driving, and contributing to the advancement of intelligent transportation systems.

Graphical Abstract

Keywords
Autonomous driving
object Detection
YOLOv8
real-time performance
intelligent transportation

Cite This Article
M,Yang & X,Fan. (2024). YOLOv8-Lite: A Lightweight Object Detection Model for Real-time Autonomous Driving Systems.IECE Transactions on Emerging Topics in Artificial Intelligence ,1(1),1-16. https://doi.org/10.62762/TETAI.2024.89422

References

[1] Armstrong Aboah, Bin Wang, Ulas Bagci, and Yaw Adu-Gyamfi. Real-time multi-class helmet violation detection using few-shot data sampling technique and yolov8. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 5349–5357, 2023.

[2] Nitika Arora, Yogesh Kumar, Rashmi Karkra, andMunish Kumar. Automatic vehicle detection systemin different environment conditions using fast r-cnn.Multimedia Tools and Applications, 81(13):18715–18735,2022.

[3] Sarthak Babbar and Jatin Bedi. Real-time traffic,accident, and potholes detection by deep learningtechniques: a modern approach for trafficmanagement. Neural Computing and Applications,35(26):19465–19479, 2023.

[4] Igor Bisio, Chiara Garibotto, Halar Haleem, FabioLavagetto, and Andrea Sciarrone. A systematic reviewof drone based road traffic monitoring system. IEEEAccess, 2022.

[5] Xumeiqi Chen. Traffic lights detection method basedon the improved yolov5 network. In 2022 IEEE 4thInternational Conference on Civil Aviation Safety andInformation Technology (ICCASIT), pages 1111–1114.IEEE, 2022.

[6] Zhichao Chen, Haoqi Guo, Jie Yang, Haining Jiao,Zhicheng Feng, Lifang Chen, and Tao Gao. Fast vehicledetection algorithm in traffic scene based on improvedssd. Measurement, 201:111655, 2022.

[7] Wenjie Du, Lianliang Chen, Haoran Wang, ZiyangShan, Zhengyang Zhou, Wenwei Li, and Yang Wang.Deciphering urban traffic impacts on air quality bydeep learning and emission inventory. journal ofenvironmental sciences, 124:745–757, 2023.

[8] Hanif Fakhrurroja, Dita Pramesti, Abdul RofiHidayatullah, Ahda Arif Fashihullisan, HarryBangkit, and Nanang Ismail. Automated license platedetection and recognition using yolov8 and ocr withtello drone camera. In 2023 International Conferenceon Computer, Control, Informatics and its Applications(IC3INA), pages 206–211. IEEE, 2023.

[9] Andreas Geiger, Philip Lenz, and Raquel Urtasun. Arewe ready for autonomous driving? the kitti visionbenchmark suite. In Conference on Computer Vision andPattern Recognition (CVPR), 2012.

[10] Marielet Guillermo, Kate Francisco, RonnieConcepcion, Arvin Fernando, Argel Bandala,Ryan Rhay Vicerra, and Elmer Dadios. A comparativestudy on satellite image analysis for road trafficdetection using yolov3-spp, keras retinanet and fullconvolutional network. In 2023 8th InternationalConference on Business and Industrial Research (ICBIR),pages 578–584. IEEE, 2023.

[11] SP Krishnendhu and Prabu Mohandas. Sad:Sensor-based anomaly detection system for smartjunctions. IEEE Sensors Journal, 2023.

[12] Songjiang Li, Shilong Wang, and Peng Wang. A smallobject detection algorithm for traffic signs based onimproved yolov7. Sensors, 23(16):7145, 2023.

[13] Xiaomei Li, Zhijiang Xie, Xiong Deng, Yanxue Wu, andYangjun Pi. Traffic sign detection based on improvedfaster r-cnn for autonomous driving. The Journal ofSupercomputing, pages 1–21, 2022.

[14] Qiuli Liu, Haixiong Ye, Shiming Wang, and ZheXu. Yolov8-cb: Dense pedestrian detection algorithmbased on in-vehicle camera. Electronics, 13(1):236,2024.

[15] Zhiqiang Liu, Jiaojiao Li, Rui Song, Chaoxiong Wu,Wei Liu, Zan Li, and Yunsong Li. Edge guided contextaggregation network for semantic segmentation ofremote sensing imagery. Remote Sensing, 14(6):1353,2022.

[16] Haoxiang Ma, Hongyu Yang, and Di Huang.Boundary guided context aggregation for semanticsegmentation. arXiv preprint arXiv:2110.14587, 2021.

[17] Usha Mittal, Priyanka Chawla, and Rajeev Tiwari.Ensemblenet: A hybrid approach for vehicle detectionand estimation of traffic density based on faster r-cnnand yolo models. Neural Computing and Applications,35(6):4755–4774, 2023.

[18] Enhao Ning, Changshuo Wang, Huang Zhang,Xin Ning, and Prayag Tiwari. Occluded personre-identification with deep learning: A surveyand perspectives. Expert Systems with Applications,239:122419, 2024.

[19] Enhao Ning, Changshuo Wang, Huang Zhang,Xin Ning, and Prayag Tiwari. Occluded personre-identification with deep learning: A surveyand perspectives. Expert Systems with Applications,239:122419, 2024.

[20] Xin Ning, Feng He, Xiaoli Dong, Weijun Li,Fayadh Alenezi, and Prayag Tiwari. Icgnet:An intensity-controllable generation network basedon covering learning for face attribute synthesis.Information Sciences, 660:120130, 2024.

[21] Xin Ning, Zaiyang Yu, Lusi Li, Weijun Li, andPrayag Tiwari. Dilf: Differentiable rendering-basedmulti-view image–language fusion for zero-shot 3dshape understanding. Information Fusion, 102:102033,2024.

[22] Mohamed Othmani. A vehicle detection andtracking method for traffic video based on faster r-cnn.Multimedia Tools and Applications, 81(20):28347–28365,2022.

[23] Karim Sattar, Feras Chikh Oughali, Khaled Assi, NedalRatrout, Arshad Jamal, and Syed Masiur Rahman.Transparent deep machine learning framework forpredicting traffic crash severity. Neural Computing andApplications, 35(2):1535–1547, 2023.

[24] Emel Soylu and Tuncay Soylu. A performancecomparison of yolov8 models for traffic sign detectionin the robotaxi-full scale autonomous vehiclecompetition. Multimedia Tools and Applications, pages1–31, 2023.

[25] Fatma M Talaat and Hanaa ZainEldin. Animproved fire detection approach based on yolo-v8for smart cities. Neural Computing and Applications,35(28):20939–20954, 2023.

[26] Devrim Unal, Ferhat Ozgur Catak, Mohammad TalalHoukan, Mohammed Mudassir, and MohammadHammoudeh. Towards robust autonomous drivingsystems through adversarial test set generation. ISAtransactions, 132:69–79, 2023.

[27] Jian Wang, Fan Li, Yi An, Xuchong Zhang, andHongbin Sun. Towards robust lidar-camera fusionin bev space via mutual deformable attention andtemporal aggregation. IEEE Transactions on Circuitsand Systems for Video Technology, 2024.

[28] Xueqiu Wang, Huanbing Gao, Zemeng Jia, and ZijianLi. Bl-yolov8: An improved road defect detectionmodel based on yolov8. Sensors, 23(20):8361, 2023.

[29] Hongyang Wei, Qianqian Zhang, Yugang Qin, XiangLi, and Yurong Qian. Yolof-f: you only look one-levelfeature fusion for traffic sign detection. The VisualComputer, pages 1–14, 2023.

[30] Jiaao Xia, Meijuan Li, Weikang Liu, and Xuebo Chen.Dsra-detr: An improved detr for multiscale traffic signdetection. Sustainability, 15(14):10862, 2023.

[31] Pengcheng Zhang, Xiaohan Yu, Xiao Bai, Chen Wang,Jin Zheng, and Xin Ning. Joint discriminativerepresentation learning for end-to-end person search.Pattern Recognition, 147:110053, 2024.

[32] Yatao Zhang, Tianhong Zhao, Song Gao, andMartin Raubal. Incorporating multimodal contextinformation into traffic speed forecasting throughgraph deep learning. International Journal ofGeographical Information Science, 37(9):1909–1935, 2023.

[33] Yanzhao Zhu and Wei Qi Yan. Traffic sign recognitionbased on deep learning. Multimedia Tools andApplications, 81(13):17779–17791, 2022.

[34] Haohao Zou, Huawei Zhan, and Linqing Zhang.Neural network based on multi-scale saliency fusionfor traffic signs detection. Sustainability, 14(24):16491,2022.


Publisher's Note
IECE stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions
IECE or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
Copyright © 2024 Institute of Emerging and Computer Engineers INC. All rights reserved.