Journal of information and communication convergence engineering

The Korea Institute of Information and Commucation Engineering (한국정보통신학회)
권호 표지
DOI QR코드

DOI QR Code

Human Detection using Real-virtual Augmented Dataset

  • Jongmin, Lee (School of IT Convergence, University of Ulsan) ;
  • Yongwan, Kim (VR/AR Content Research Section, Communications & Media Research Laboratory, Electronics and Telecommunications Research Institute (ETRI)) ;
  • Jinsung, Choi (VR/AR Content Research Section, Communications & Media Research Laboratory, Electronics and Telecommunications Research Institute (ETRI)) ;
  • Ki-Hong, Kim (VR/AR Content Research Section, Communications & Media Research Laboratory, Electronics and Telecommunications Research Institute (ETRI)) ;
  • Daehwan, Kim (School of IT Convergence, University of Ulsan)
  • Received : 2022.12.04
  • Accepted : 2023.02.17
  • Published : 2023.03.31
Download PDF
⟨ Previous Next ⟩

Abstract

This paper presents a study on how augmenting semi-synthetic image data improves the performance of human detection algorithms. In the field of object detection, securing a high-quality data set plays the most important role in training deep learning algorithms. Recently, the acquisition of real image data has become time consuming and expensive; therefore, research using synthesized data has been conducted. Synthetic data haves the advantage of being able to generate a vast amount of data and accurately label it. However, the utility of synthetic data in human detection has not yet been demonstrated. Therefore, we use You Only Look Once (YOLO), the object detection algorithm most commonly used, to experimentally analyze the effect of synthetic data augmentation on human detection performance. As a result of training YOLO using the Penn-Fudan dataset, it was shown that the YOLO network model trained on a dataset augmented with synthetic data provided high-performance results in terms of the Precision-Recall Curve and F1-Confidence Curve.

Keywords

Acknowledgement

This research was supported by Culture, Sports and Tourism R&D Program through the Korea Creative Content Agency grant funded by the Ministry of Culture, Sprots and Tourism in 2022 (Project Name: Development of Virtual Reality Performance Platform Supporting Multiuser Participation and Realtime Interaction, Project Number: R2021040046, Contribution Rate: 100%)

References

  1. I. Joshi, M. Grimmer, C. Rathgeb, C. Busch, F. Bremond, and A. Dantcheva, "Synthetic data in human analysis: A survey," arXiv preprint arXiv:2208.09191, Aug. 2022. DOI: 10.48550/arXiv.2208.09191.
  2. H, Kim, D, Kim, J, Kim, and S, Im, "Data augmentation scheme for semi-supervised video object segmentation," Journal of Broadcast Engineering, vol. 27, no. 1, pp. 13-19, 2022.
  3. K. Man and J. Chachi, "A Review of Synthetic Image Data and Its Use in Computer Vision," Journal of Imaging, vol. 8, no.11, p. 810, Nov. 2022. DOI: 10.3390/jimaging8110310.
  4. G. Varol, J. Romero, X. Martin, N. Mahmood, M. J. Black, I. Laptev, and C. Schmid, "Learning from synthetic humans," in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, USA. pp. 109-117, 2017. DOI: 10.1109/CVPR.2017.492.
  5. J. Mu, W. Qiu, G. D. Hanger, and A. L. Yuille, "Learning from synthetic animals," in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Seattle, USA, pp. 12386-12395, 2020. DOI: 10.1109/CVPR42600.2020.01240.
  6. Q. Wang, J. Gao, W. Lin, and Y. Yuan, "Pixel-wise crowd understanding via synthetic data," International Journal of Computer Vision, vol. 129, no. 1, pp. 225-245, Jan. 2021. DOI: 10.1007/s11263-020-01365-4.
  7. Unity. [Online] Available: https://www.unity.com.
  8. Penn-Fudan Database for Pedestrian Detection and Segmentation. [Online] Available: https://www.cis.upenn.edu/~jshi/ped_html/.
  9. Remove any unwanted object, defect, people of text you're your pictures in seconds, cleanup.pictures. [Online] Available: https://cleanup.pictures/
  10. G. Jocher, Code. [Online] Available: https://github.com/ultralytics/yolov5.
  11. C. Y. Wang, H. Y. M. Liao, Y. H. Wu, P. Y. Chen, J. W. Hsieh, and I. H. Yeh, "CSPNet: A new backbone that can enhance learning capability of CNN," in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, Seattle, USA, pp. 390-391, 2020. DOI: 10.1109/CVPRW50498.2020.00203.
  12. Z, Tian, C, Shen, H. Chen, and T, He, "FCOS: full convolutional one-stage object detection," in Proceedings of the IEEE International Conference on Computer Vision, Seoul, Korea, pp. 9627-9636, 2019. DOI: 10.1109/ICCV.2019.00972.
  13. Pedestrian Detection Data set, Kaggle. [Online] Available: https://www.kaggle.com/datasets/karthika95/pedestrian-detection.