CLIENT-SERVER SYSTEM FOR PEOPLE IDENTIFICATION AND TRACKING IN INDOOR BASED ON VIDEO DATA ANALYSIS
Article Sidebar
Main Article Content
Abstract
In this paper we present software that was developed for identification and tracking of multiple people
indoors. Proposed approach based on the analysis of image sequences obtained from stationary CCTV cameras.
Person detection and tracking procedures use convolutional neural network architectures. Tracking algorithm
is characterized by collecting face recognition results for the correct assignment of a person's external features
in different frames. An integral descriptor defines each person and contains convolutional neural network based
face features and person’s image features. His structure allows us to track people when faces can’t be recognized
properly. The software implementation uses the OpenCV library for basic image processing operations. For the
main procedures of our algorithm we use software and hardware architecture for parallel computing that is implemented on an Nvidia GPU and CUDA technology. The proposed approach achieves real-time processing
if there are five or less people in a frame simultaneously. The computational experiments are conducted on a PC
with GPU NVIDIA GTX 1060 and CPU Intel i7-5820k. The output data is processed image sequence with rectangular areas of people’s location and their index. Overall number of people in the current image is showed as well.
The developed client-server system was tested using a personal computer and mobile devices running on Android
and iOS operating systems.
Article Details
This work is licensed under a Creative Commons Attribution 4.0 International License.
References
Chahyati, D. Tracking People by Detection Using CNN Features / D. Chahyati, M.I. Fanany, A.M. Arymurthy // In Proceedings of the 4th Information Systems International Conference (ISICO 2017). – 2017. – P. 167–172.
Wojke, N. Simple online and realtime tracking with a deep association metric / N. Wojke, A. Bewley, D. Paulus // In Proceedings of the IEEE International Conference on Image Processing (ICIP). – 2017. – P. 3645–3649.
Bohush, R. Robust Person Tracking Algorithm Based on Convolutional Neural Network for Indoor Video Surveillance / R. Bohush, I. Zakharava // Communications in Computer and Information Science. – 2019. – Vol. 1055. – P. 289–300.
YOLOv4: Optimal Speed and Accuracy of Object Detection [Electronic resource]. – Mode of access: https://arxiv.org/pdf/2004.10934.pdf. – Date of access: 29.09.2020.
Ma M.H. Multi-View Face Detection and Landmark Localization Based on MTCNN// M.H. Ma, J. Wang // In Proceedings of the 2018 Chinese Automation Congress (CAC). – 2018. – P. 4200–4205. – DOI:10.1109/cac.2018.8623535.
WIDER FACE: A Face Detection Benchmark [Electronic resource] // Sh. Yang [et al.] // Computing Research Repository ; 2015 arXiv:1511.06523. – Mode of access: https://arxiv.org/pdf/1511.06523.pdf. – (дата обращения 16.06.2019).
RetinaFace: Single-stage Dense Face Localisation in the Wild [Electronic resource] // J. Deng [et al.] // Computing Research Repository ; 2019 arXiv:1905.00641v2. – Mode of access: https://arxiv.org/pdf/1905.00641.pdf. – Date of access: 16.06.2019.
InsightFace [Electronic resource]. – Mode of access: https://github.com/deepinsight/insightface/wiki/Model-ZooMTCNN. – Date of access: 29.09.2020.
Богуш, Р.П. Алгоритм сопровождения людей на видеопоследовательностях с использованием сверточных нейронных сетей для видеонаблюдения внутри помещений / Р.П. Богуш, И.Ю. Захарова // Компьютерная оптика. – 2020. – Т. 44, № 1. – С. 109–116. – DOI: 10.18287/2412-6179-CO-565.