Proceedings of International Conference on Applied Innovation in IT · 2021/04/28 · Vol. 9 · Issue 1 · pp. 61–67
Urban Environment Simulator for Train Data Generation Toward CV Object RecognitionModel of Damage Accumulation in Predicting the Technical Condition of a Fiber-Optic Cable
Kirill Karpov, Ivan Luzianin, Maksim Iushchenko, Eduard Siemens
Detecting moving objects in an urban environment is a challenging and widely explored problem in computer vision. This task requires huge amounts of data. Their obtaining and labeling is challenging. However the available datasets are not always fit the task. This work proposes a framework for synthesizing the train data based on 3D visualization of an urban environment using Unity 3D. Methods of mathematical statistics and distribution theory were used to build the background models of the framework. The framework, presented in the article, allows to simulate the real urban environment in an adjustable 3D virtual scene. It considers differ- ent environmental parameters amd makes possible to simulate the real behavior and physical characteristics of moving objects.
Keywords
Virtual RealityComputer VisionObject DetectionPublic Street EnvironmentSimulationTrafficPedes- trianDistributionLabeling
References
[I.Matveev,K.Karpov,A.Yurchenko,andE.Siemens, “The object tracking algorithm using dimensional based detection for public street environment,”Eurasian Physical Technical Journal, vol. 17, pp. 123– 127, Dec. 2020.
I. Matveev, K. Karpov, I. Chmielewski, E. Siemens, and A. Yurchenko, “Fast object detection using di- mensional based features for public street environ- ments,” Smart Cities, vol. 3, no. 1, pp. 93–111, 2020.
X.Ouyang,Y.Cheng,Y.Jiang,C.-L.Li,andP.Zhou, “Pedestrian-synthesis-gan: Generating pedestrian data in real scene and beyond,” arXiv preprint arXiv:1804.02047, 2018.
W. Zhang, K. Wang, H. Qu, J. Zhao, and F.-Y. Wang, “Scene-specific pedestrian detection based on parallel vision,” arXiv preprint arXiv:1712.08745, 2017.
H. Hattori, V. N. Boddeti, K. Kitani, and T. Kanade, “Learning scene-specific pedestrian detectors without real data,” in 2015 IEEE Conference on Computer Vi- sion and Pattern Recognition (CVPR). Boston, MA, USA: IEEE, Jun. 2015, pp. 3819–3827.
J. Nilsson, P. Andersson, I. Y. Gu, and J. Fredriks- son, “Pedestrian detection using augmented training data,” in 2014 22nd International Conference on Pat- tern Recognition. IEEE, 2014, pp. 4548–4553.
F. Moretti, S. Pizzuti, S. Panzieri, and M. Annunziato, “Urban traffic flow forecasting through statistical and neural network bagging ensemble hybrid modeling,” Neurocomputing, vol. 167, pp. 3–7, 2015.
L. Bartusˇka, V. Biba, and R. Kampf, “Modeling of daily traffic volumes on urban roads,” 2016.
Metropolitan Washington Council of Govern- ments. Traffic Counts - Hourly Classification Counts 2017. Accessed Mar. 20, 2021. [Online]. Available: https://rtdc-mwcog.opendata.arcgis.com/ datasets/fae4f4ebf99c45088adbfba504efd650
Bike Counts (Eco Counter). City of Edmonton. Accessed Mar. 20, 2021. [Online]. Available: https: //data.edmonton.ca/Monitoring-and-Data-Collection/ Bike-Counts-Eco-Counter-/tq23-qn4m
City of Melbourne Open Data Team. Pedes- trian Counting System - Monthly (counts per hour). Accessed Mar. 20, 2021. [Online]. Avail- able: https://data.melbourne.vic.gov.au/Transport/ Pedestrian-Counting-System-Monthly-counts-per-hour/ b2ak- trbp
S. Buchmueller and U. Weidmann, “Parameters of pedestrians, pedestrian traffic and walking facilities,” IVT Schriftenreihe, vol. 132, 2006.
E. Brolin, “Anthropometric diversity and considera- tion of human capabilities,” p. 101.
N. HajiGhassemi and M. Deisenroth, “Analytic long- term forecasting with periodic gaussian processes,” in Artificial Intelligence and Statistics. pp. 303–311.
·
This work is licensed under a
Creative Commons Attribution-ShareAlike 4.0 International License
