Generation of Pedestrian Crossing Scenarios Using Ped-Cross Generative Adversarial Network
Abstract
:1. Introduction
2. Related Work
2.1. Pedestrian Deaths and CAVs
2.2. Pedestrian Datasets
2.3. Human Pose Estimation
3. Pedestrian Scenario Dataset
3.1. Dataset Curation
3.1.1. Additional Labeling
3.1.2. Pose Dataset
- Keypoints for the shoulders to be in the top 30% of the prediction;
- Keypoints for the hips should be within the 40% to 60% range in the prediction;
- Keypoints for the feet should be in the bottom 40%;
- Keypoints for the feet should not be above the knee; and
- Keypoints for both knees should be within 10% height of each other.
3.2. Dataset Results
3.2.1. Image and Pose datasets
3.2.2. Labels and Classes
4. Generative Adversarial Networks
5. Ped-Cross GAN
5.1. Network Architecture
5.2. Training
5.3. Hyperparameters
6. Ped-Cross GAN Results
6.1. Validation Method
6.2. Validation Results
6.2.1. Normal Validation
6.2.2. Reverse Validation
6.3. Visual Results
7. Discussion
7.1. Normal Validation
7.2. Reverse Validation
7.3. Training of Ped-Cross GAN
8. Conclusions and Future Work
Author Contributions
Funding
Institutional Review Board Statement
Informed Consent Statement
Data Availability Statement
Conflicts of Interest
References
- World Health Organization. Road Traffic Injuries. Available online: https://www.who.int/news-room/fact-sheets/detail/road-traffic-injuries (accessed on 5 January 2021).
- Singh, S. Critical reasons for crashes investigated in the National Motor Vehicle Crash Causation Survey; National Highway Traffic Safety Administration: Washington, DC, USA, 2015; pp. 1–2. Available online: https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/812115 (accessed on 5 January 2021).
- Combs, T.S.; Sandt, L.S.; Clamann, M.P.; McDonald, N.C. Automated Vehicles and Pedestrian Safety: Exploring the Promise and Limits of Pedestrian Detection. Am. J. Prev. Med. 2019, 56, 1–7. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Griggs, T.; Wakabayashi, D. How a Self-Driving Uber Killed a Pedestrian in Arizona. New York Times, 2018. [Google Scholar]
- Levin, S.; Wong, J.C. Self-Driving Uber Kills Arizona Woman in First Fatal Crash Involving Pedestrian. The Gaurdian, 2018. [Google Scholar]
- Bradshaw, T. Self-Driving Cars under Scrutiny after Uber Pedestrian Death. Financial Times, 2018. [Google Scholar]
- Dollár, P.; Wojek, C.; Schiele, B.; Perona, P. Pedestrian detection: A benchmark. In Proceedings of the 2009 IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops, CVPR Workshops 2009, Miami, FL, USA, 20–25 June 2009; pp. 304–311. [Google Scholar] [CrossRef]
- World Health Organization. Global Status Report on Road Safety—2015; Technical Report; World Health Organisation: Geneva, Switzerland, 2005; ISBN 9789241564854. [Google Scholar]
- European Commission. Road safety in the European Union—Trends, statistics and main challenges. Mobil. Transp. 2015, 1–24. [Google Scholar] [CrossRef]
- National Highway Traffic Safety Administration (NHTSA). Traffic Safety Facts—Pedestrians; 2018. Available online: https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/812493 (accessed on 5 January 2021).
- Reynolds, S.; Tranter, M.; Baden, P.; Mais, D.; Dhani, A.; Wolch, E.; Bhagat, A. Reported Road Casulatites Great Britain: 2016; Technical Report September; Department for Transport: London, UK, 2007. [Google Scholar]
- Dalal, N.; Triggs, W. Histograms of Oriented Gradients for Human Detection. In Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition CVPR05, San Diego, CA, USA, 20–25 June 2005; Volume 1, pp. 886–893. [Google Scholar] [CrossRef] [Green Version]
- Kotseruba, I.; Rasouli, A.; Tsotsos, J.K. Joint Attention in Autonomous Driving (JAAD). arXiv 2016, arXiv:1609.04741. [Google Scholar]
- Schneider, N.; Gavrila, D.M. Pedestrian path prediction with recursive Bayesian filters: A comparative study. Lect. Notes Comput. Sci. 2013, 8142 LNCS, 174–183. [Google Scholar] [CrossRef] [Green Version]
- Zhang, L.; Lin, L.; Liang, X.; He, K. Is faster R-CNN doing well for pedestrian detection? Lect. Notes Comput. Sci. 2016, 9906 LNCS, 443–457. [Google Scholar] [CrossRef] [Green Version]
- Li, J.; Liang, X.; Shen, S.; Xu, T.; Feng, J.; Yan, S. Scale-Aware Fast R-CNN for Pedestrian Detection. IEEE Trans. Multimed. 2018, 20, 985–996. [Google Scholar] [CrossRef] [Green Version]
- Du, X.; El-Khamy, M.; Morariu, V.I.; Lee, J.; Davis, L. Fused Deep Neural Networks for Efficient Pedestrian Detection. arXiv 2018, arXiv:1805.08688. [Google Scholar]
- Rasouli, A.; Kotseruba, I.; Tsotsos, J.K. Are They Going to Cross? A Benchmark Dataset and Baseline for Pedestrian Crosswalk Behavior. In Proceedings of the 2017 IEEE International Conference on Computer Vision Workshops, ICCVW 2017, Venice, Italy, 22–29 October 2017; pp. 206–213. [Google Scholar] [CrossRef]
- Cao, Z.; Hidalgo, G.; Simon, T.; Wei, S.e.; Sheikh, Y. OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields. IEEE Trans. Pattern Anal. Mach. Intell. 2019, 43, 172–186. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Fang, H.s.; Xie, S.; Tai, Y.w.; Lu, C. RMPE: Regional Multi-person Pose Estimation. In Proceedings of the 2017 IEEE International Conference on Computer Vision (ICCV), Venice, Italy, 22–29 October 2017; pp. 2353–2362. [Google Scholar] [CrossRef] [Green Version]
- Lin, T.Y.; Maire, M.; Belongie, S.; Hays, J.; Perona, P.; Ramanan, D.; Dollár, P.; Zitnick, C.L. Microsoft COCO: Common objects in context. Lect. Notes Comput. Sci. 2014, 8693 LNCS, 740–755. [Google Scholar] [CrossRef] [Green Version]
- Andriluka, M.; Pishchulin, L.; Gehler, P.; Schiele, B. 2D human pose estimation: New benchmark and state of the art analysis. In Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Columbus, OH, USA, 23–28 June 2014; pp. 3686–3693. [Google Scholar] [CrossRef]
- Goodfellow, I.J.; Pouget-Abadie, J.; Mirza, M.; Xu, B.; Warde-Farley, D.; Ozair, S.; Courville, A.; Bengio, Y. Generative Adversarial Networks. Neural Inf. Process. Syst. 2014, 27, 2672–2680. [Google Scholar]
- Nash, J.F. Equilibrium points in n-person games. Proc. Natl. Acad. Sci. USA 1950, 36, 48–49. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Fedus, W.; Rosca, M.; Lakshminarayanan, B.; Dai, A.M.; Mohamed, S.; Goodfellow, I. Many Paths to Equilibrium: GANs Do Not Need to Decrease a Divergence At Every Step. arXiv 2017, arXiv:1710.08446. [Google Scholar]
- Arjovsky, M.; Chintala, S.; Bottou, L. Wasserstein GAN. arXiv 2017, arXiv:1701.07875. [Google Scholar]
- Spooner, J.; Cheah, M.; Palade, V.; Kanarachos, S.; Daneshkhah, A. Generation of pedestrian pose structures using generative adversarial networks. In Proceedings of the 8th IEEE International Conference on Machine Learning and Applications, ICMLA 2019, Boca Raton, FL, USA, 16–19 December 2019. [Google Scholar] [CrossRef]
- Hochreiter, S.; Schmidhuber, J. Long short-term memory. Neural Comput. 1997, 9, 1735–1780. [Google Scholar] [CrossRef]
- Lin, X.; Amer, M.R. Human Motion Modeling using DVGANs. arXiv 2018, arXiv:1804.10652. [Google Scholar]
- Kingma, D.P.; Ba, J. Adam: A Method for Stochastic Optimization. arXiv 2014, arXiv:1412.6980. [Google Scholar]
- Hinton, G.; Srivastava, N.; Swersky, K. Lecture 6e: Neural Networks for Machine Learning. Coursera 2014. [Google Scholar] [CrossRef] [Green Version]
World Region | Proportion |
---|---|
Africa | 39% |
Eastern Mediterranean | 27% |
Europe | 26% |
Western Pacific | 23% |
The Americas | 22% |
South-East Asia | 13% |
World | 22% |
City, State | Proportion | Total Road Deaths |
---|---|---|
New York, NY | 59.6% | 230 |
San Francisco, CA | 50.0% | 28 |
Boston, MA | 48.1% | 27 |
Fresno, CA | 46.2% | 13 |
San Diego, CA | 43.8% | 96 |
Philadelphia, PA | 42.6% | 101 |
Los Angeles, CA | 41.3% | 315 |
Number | Label |
---|---|
1 | Crossing from the left |
2 | Diagonal towards cross left |
3 | Diagonal adjacent cross left |
4 | Crossing from the right |
5 | Diagonal towards cross right |
6 | Diagonal adjacent cross right |
7 | Walk towards traffic, no cross |
8 | Walk adjacent to traffic, no cross |
9 | Stand left |
10 | Stand right |
Action | Label |
---|---|
Speed | No Movement, Slow walk, |
Walk, Jog, Run | |
Hesitation | Yes, No |
Peeking | Yes, No |
Looking | Yes, No |
Distraction | Yes, No |
Waiting | Yes, No |
Waving | Yes, No |
Jump back | Yes, No |
Intoxicated | Yes, No |
Freeze | Yes, No |
Trip | Yes, No |
Other mobility | Skateboard, Rollerblade, |
Scooter, Other, N/A |
Desciptive | Label |
---|---|
Age Range | 0–15, 15–60, 60+ |
Gender | Male, Female, Unknown |
Ethnicity | White, Asian, Black, |
Mixed race, Unknown | |
Occluded | Yes, No |
Occluded by ped | Yes, No |
Full body | Yes, No |
Hunched over | Yes, No |
With object | Shopping, Dog, Pram, |
Crutches, Walking frame, | |
Suitcase, Other |
Speed | |||||||
---|---|---|---|---|---|---|---|
0 | 1 | 2 | 3 | 4 | Total | ||
Movement Class | 1 | 0 | 14 | 170 | 22 | 0 | 206 |
2 | 0 | 3 | 42 | 3 | 0 | 48 | |
3 | 0 | 5 | 28 | 4 | 2 | 39 | |
4 | 0 | 18 | 226 | 21 | 6 | 271 | |
5 | 0 | 6 | 28 | 5 | 0 | 39 | |
6 | 0 | 5 | 54 | 4 | 0 | 63 | |
7 | 0 | 8 | 64 | 0 | 1 | 73 | |
8 | 0 | 20 | 56 | 1 | 0 | 77 | |
9 | 20 | 11 | 1 | 0 | 0 | 32 | |
10 | 50 | 31 | 3 | 0 | 0 | 84 | |
Total | 70 | 121 | 672 | 60 | 9 | 932 |
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |
© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
Share and Cite
Spooner, J.; Palade, V.; Cheah, M.; Kanarachos, S.; Daneshkhah, A. Generation of Pedestrian Crossing Scenarios Using Ped-Cross Generative Adversarial Network. Appl. Sci. 2021, 11, 471. https://doi.org/10.3390/app11020471
Spooner J, Palade V, Cheah M, Kanarachos S, Daneshkhah A. Generation of Pedestrian Crossing Scenarios Using Ped-Cross Generative Adversarial Network. Applied Sciences. 2021; 11(2):471. https://doi.org/10.3390/app11020471
Chicago/Turabian StyleSpooner, James, Vasile Palade, Madeline Cheah, Stratis Kanarachos, and Alireza Daneshkhah. 2021. "Generation of Pedestrian Crossing Scenarios Using Ped-Cross Generative Adversarial Network" Applied Sciences 11, no. 2: 471. https://doi.org/10.3390/app11020471
APA StyleSpooner, J., Palade, V., Cheah, M., Kanarachos, S., & Daneshkhah, A. (2021). Generation of Pedestrian Crossing Scenarios Using Ped-Cross Generative Adversarial Network. Applied Sciences, 11(2), 471. https://doi.org/10.3390/app11020471