Hướng tới phát hiện sự phân tâm của người lái xe: một phương pháp học liên bang bảo vệ quyền riêng tư

Wenguang Zhou1, Zhiwei Jia1, Chao Feng1, Huali Lu2, Feng Lyu2, Ling Li1
1State Key Laboratory of Disaster Prevention & Reduction for Power Grid, Changsha University of Science and Technology, Changsha, China
2School of Computer Science and Engineering, Central South University, Changsha, China

Tóm tắt

Sự phân tâm của người lái xe là nguyên nhân chính gây ra tai nạn giao thông, và hệ thống tự động phát hiện sự phân tâm của người lái xe có thể giảm thiểu đáng kể các sự cố không may này. Số lượng mẫu hợp lệ đủ lớn là điều kiện tiên quyết để thiết kế một hệ thống như vậy, và việc bảo vệ quyền riêng tư là điều kiện ràng buộc để sử dụng những mẫu này. Để giải quyết vấn đề này, một thuật toán học liên bang bảo vệ quyền riêng tư cho việc phát hiện sự phân tâm của người lái xe trong quá trình lái xe được đề xuất. Để ngăn chặn tình trạng rò rỉ gradient của mô hình toàn cầu trong quá trình huấn luyện và rò rỉ dữ liệu nhạy cảm, một cơ chế cắt gradient thích ứng (AGC) được giới thiệu nhằm bảo vệ quyền riêng tư trong khi giảm bớt tải trọng truyền thông. Ngoài ra, một kỹ thuật riêng tư vi sai với cơ chế Gaussian được áp dụng để bảo vệ thêm quyền riêng tư của những người tham gia. Tiếp theo, một thành phần ràng buộc được thêm vào hàm mục tiêu để đảm bảo sự tương đồng giữa các mô hình địa phương và toàn cầu, cũng như cải thiện tốc độ hội tụ của mô hình. Để khắc phục các vấn đề về nền tảng đa dạng và các mục tiêu nhỏ trong việc phát hiện sự phân tâm của người lái xe, một hàm mất mát hộp giới hạn SIoU và cơ chế tự chú ý được sử dụng trong bộ phát hiện YOLOv5 cải tiến để thúc đẩy hiệu suất của mô hình. Cuối cùng, hiệu quả của thuật toán được xác thực thông qua các thí nghiệm trên các tập dữ liệu thực tế. So với mô hình huấn luyện dữ liệu tập trung, bộ phát hiện YOLOv5 cải tiến cải thiện hiệu suất phát hiện sự phân tâm của người lái xe với độ chính xác và độ tin cậy cao hơn.

Từ khóa

#phát hiện sự phân tâm của người lái xe #học liên bang #bảo vệ quyền riêng tư #cắt gradient thích ứng #YOLOv5 #hiệu suất mô hình

Tài liệu tham khảo

Guo F, Klauer SG, Fang Y, Hankey JM, Antin JF, Perez MA, Lee SE, Dingus TA (2017) The effects of age on crash risk associated with driver distraction. International journal of epidemiology 46(1):258–265 Aksjonov A, Nedoma P, Vodovozov V, Petlenkov E, Herrmann M (2018) Detection and evaluation of driver distraction using machine learning and fuzzy logic. IEEE Transactions on Intelligent Transportation Systems 20(6):2048–2059 Cunningham ML, Regan MA (2018) Driver distraction and inattention in the realm of automated driving. IET Intelligent Transport Systems 12(6):407–413 Zhang C, Eskandarian A (2021) A Survey and Tutorial of EEG-Based Brain Monitoring for Driver State Analysis. IEEE/CAA J Autom Sin 8(7):1222–1242 Sun Y, Yu X (2014) An innovative nonintrusive driver assistance system for vital signal monitoring. IEEE journal of biomedical and health informatics 18(6):1932–1939 Daza IG, Bergasa LM, Bronte S, Yebes JJ, Almazán J, Arroyo R (2014) Fusion of optimized indicators from advanced driver assistance systems (adas) for driver drowsiness detection. Sensors 14(1):1106–1131 Deshmukh SV, Dehzangi O (2017) Ecg-based driver distraction identification using wavelet packet transform and discriminative kernel-based features. In: 2017 IEEE International Conference on Smart Computing (SMARTCOMP), p 1–7. IEEE, Hong Kong Dong Y, Hu Z, Uchimura K, Murayama N (2010) Driver inattention monitoring system for intelligent vehicles: A review. IEEE transactions on intelligent transportation systems 12(2):596–614 Peng Y, Boyle LN, Hallmark SL (2013) Driver’s lane keeping ability with eyes off road: Insights from a naturalistic study. Accident Analysis & Prevention 50(2):628–634 Seshadri K, Juefei-Xu F, Pal DK, Savvides M, Thor CP (2015) Driver cell phone usage detection on strategic highway research program (shrp2) face view videos. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, p 35–43 Alam L, Hoque MM (2019) Real-time distraction detection based on driver’s visual features. 2019 International Conference on Electrical, vol 5. Computer and Communication Engineering (ECCE). IEEE, New York, p 1–6 Moslemi N, Soryani M, Azmi R (2021) Computer vision-based recognition of driver distraction: A review. Concurrency and Computation: Practice and Experience 33(24):6475 McMahan B, Moore E, Ramage D, Hampson S, y Arcas BA (2017) Communication-efficient learning of deep networks from decentralized data. In: Singh, A., Zhu, J. (eds.) Artificial Intelligence and Statistics, vol. 54, p 1273–1282. PMLR, New York Xu J, Glicksberg BS, Su C, Walker P, Bian J, Wang F (2021) Federated learning for healthcare informatics. Journal of Healthcare Informatics Research 5(2):1–19 Long G, Shen T, Tan Y, Gerrard L, Clarke A, Jiang J (2021) Federated learning for privacy-preserving open innovation future on digital health. Humanity Driven AI: Productivity. Well-being, Sustainability and Partnership. Springer, New York, p 113–133 Kumar R, Khan AA, Kumar J, Golilarz NA, Zhang S, Ting Y, Zheng C, Wang W et al (2021) Blockchain-federated-learning and deep learning models for covid-19 detection using ct imaging. IEEE Sensors Journal 21(14):16301–16314 Posner J, Tseng L, Aloqaily M, Jararweh Y (2021) Federated learning in vehicular networks: Opportunities and solutions. IEEE Network 35(2):152–159 Duan S, Wang D, Ren J, Lyu F, Zhang Y, Wu H, Shen X (2023) Distributed artificial intelligence empowered by end-edge-cloud computing: A survey. IEEE Communications Surveys & Tutorials 25(1):591–624 Zafar A, Prehofer C, Cheng C-H (2021) Federated learning for driver status monitoring. In: 2021 IEEE International Intelligent Transportation Systems Conference (ITSC), p 1463–1469. IEEE, Singapore Zhang L, Saito H, Yang L, Wu J (2022) Privacy-preserving federated transfer learning for driver drowsiness detection. IEEE Access 10(2):80565–80574 Bonawitz K, Ivanov V, Kreuter B, Marcedone A, McMahan HB, Patel S, Ramage D, Segal A, Seth K (2017) Practical secure aggregation for privacy-preserving machine learning. In: Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security, p 1175–1191. Association for Computing Machinery, Dallas, Texas, USA Ghazi B, Pagh R, Velingker A (2021) Scalable and differentially private distributed aggregation in the shuffled model. Int Conf Mach Learn 6(8):32–48 Deng Y, Lyu F, Ren J, Chen Y, Yang P, Zhou Y, Zhang Y (2022) Improving federated learning with quality-aware user incentive and auto-weighted model aggregation. IEEE Transactions on Parallel and Distributed Systems 33(12):4515–4529 Deng Y, Lyu F, Ren J, Zhang Y, Zhou Y, Zhang Y, Yang Y (2021) Share: Shaping data distribution at edge for communication-efficient hierarchical federated learning. In: 2021 IEEE 41st International Conference on Distributed Computing Systems (ICDCS) 23:24–34 Geyer RC, Klein T, Nabi M (2017) Differentially private federated learning. A Client Level Perspective 12(2):157–175 Bhowmick A, Duchi J, Freudiger J, Kapoor G, Rogers R (2018) Protection against reconstruction and its applications in private federated learning. Mach Learn 76(2):812–824 Agarwal N, Suresh AT, Yu FXX, Kumar S, McMahan B (2018) cpsgd: Communication-efficient and differentially-private distributed sgd. Adv Neural Inf Process Syst 31(2):7575–7586 Reddi S, Charles Z, Zaheer M, Garrett Z, Rush K, Koneˇcn`y J, Kumar S, McMahan HB, (2021) Adaptive federated optimization. Int Conf Learn Represent 8(3):295–323 Ding X, Wang C, Choo K-KR, Jin H (2019) A novel privacy preserving framework for large scale graph data publishing. IEEE transactions on knowledge and data engineering 33(2):331–343 Deng Y, Lyu F, Ren J, Wu H, Zhou Y, Zhang Y, Shen X (2021) Auction: Automated and quality-aware client selection framework for efficient federated learning. IEEE Transactions on Parallel and Distributed Systems 33(8):1996–2009 Li T, Sahu AK, Zaheer M, Sanjabi M, Talwalkar A, Smith V (2020) Federated optimization in heterogeneous networks. Proceedings of Machine learning and systems 2(2):429–450 Jocher G, Chaurasia A, Borovec J (2021) ultralytics/yolov5: v6.0 - YOLOv5n ’Nano’ models. Roboflow Integ OpenCV DNN Supp 28(7):657-671 Gevorgyan Z (2019) Siou loss: More powerful learning for bounding box regression. 2019 IEEE/CVF Conf Comput Vis Pattern Recognit (CVPR) 89(2):658–666 http://arxiv.org/abs/2205.12740 Hou Q, Zhou D, Feng J (2021) Coordinate attention for efficient mobile network design. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, p 13713–13722. IEEE, Nashville, TN, USA Rezatofighi H, Tsoi N, Gwak J, Sadeghian A, Reid I, Savarese S (2019) Generalized intersection over union: A metric and a loss for bounding box regression. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, p 658–666. IEEE, Salt Lake City, UT, USA Montoya A, Holman D, Smith T, Kan W (2016) State farm distracted driver detection. Publish at https://kaggle.com/competitions/state-farm-distracted-driver-detection Redmon J, Farhadi A (2016) You Only Look Once: Unified, Real-Time Object Detection. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 36(2):779–788 Ren S, He K, Girshick R, Sun J (2015) Faster r-cnn: Towards real-time object detection with region proposal networks. Advances in neural information processing systems 28(2):152–161 Redmon J, Farhadi A (2018) Yolov3: An incremental improvement 36(2):78–84. arXiv preprint arXiv:1804.02767 Wang C-Y, Bochkovskiy A, Liao H-YM (2023) Yolov7: Trainable bag-offreebies sets new state-of-the-art for real-time object detectors. Proc IEEE/CVF Conf Comput Vis Pattern Recognit 34(12):7464–7475 Jocher G, Chaurasia A, Qiu J (2023) YOLO by Ultralytics, 8.0.0. Publish at https://github.com/ultralytics/ultralytics Carion N, Massa F, Synnaeve G, Usunier N, Kirillov A, Zagoruyko S (2020) End-to-end object detection with transformers. In: Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part I 16, p 213–229. Springer