Optimization of parallel computing techniques for real-time hazard and obstacle detection in assistive systems for visually impaired individuals

Mukhriddin Mukhiddinov; Javokhir Kurbonov

doi:10.65112/tcmis.10089

Authors

Mukhriddin Mukhiddinov Department of Computer Systems, Tashkent University of Information Technologies named after Muhammad al-Khwarizmi, Tashkent 100200, Uzbekistan & Department of Industrial Management and Digital Technologies, Nordic International University, Tashkent 100128, Uzbekistan https://orcid.org/0000-0002-1424-0799
Javokhir Kurbonov Department of Computer Systems, Tashkent University of Information Technologies named after Muhammad al-Khwarizmi, Tashkent 100200, Uzbekistan https://orcid.org/0009-0001-2892-5625

DOI:

https://doi.org/10.65112/tcmis.10089

Keywords:

computer vision, artificial intelligence, object detection, real-time systems, assistive technologies for the visually impaired, YOLO, depth estimation

Abstract

This article addresses the optimization of real-time hazard and obstacle detection for visually impaired individuals through parallel computing. An "AI Blind Assist System" is developed that integrates the YOLOv8 model for object detection, the MiDaS model for monocular relative depth estimation, and the pyttsx3 library for voice-based alerts, combined with a distance-based three-tier risk classification (high risk < 2 m, medium risk 2-5 m, safe > 5 m). YOLOv8 detection and MiDaS depth estimation were executed as two concurrent tasks using Python's multiprocessing and threading modules on a system equipped with an 11th-generation Intel Core i5 CPU (6 cores/12 threads), 16 GB RAM, and an NVIDIA GeForce RTX 2050 GPU (4 GB VRAM). Compared with sequential execution, the parallel implementation reduced the measured end-to-end processing time per frame from 0.286 s to 0.210 s, corresponding to a 1.36× latency speedup. At the same time, the processing throughput increased from 3.5 to 5.5 frames per second (FPS), corresponding to a 1.57× throughput speedup. This difference is attributable to pipeline overlap between the concurrently executed YOLOv8 and MiDaS tasks. Both figures indicate a sub-linear speedup relative to the two execution units employed. The obtained results demonstrate that task-level parallel computing measurably improves the responsiveness of real-time assistive navigation systems for visually impaired individuals.

Downloads

Download data is not yet available.

References

[1] World Health Organization, World Report on Vision. Geneva, Switzerland: World Health Organization, 2019.

[2] K. Potdar, C. Pai, and S. Akolkar, “A convolutional neural network based live object recognition system as blind aid,” Int. J. Comput. Appl., 2018.

[3] S. Chandra, U. Sharma, and D. Khilnani, “Computer vision and depth sensor-powered smart cane for real-time obstacle detection,” 2025.

[4] M. J. Shafiee, B. Chywl, F. Li, and A. Wong, “Fast YOLO: A fast You Only Look Once system for real-time embedded object detection in video,” 2017. 10.15353/vsnl.v3i1.171

[5] R. Szeliski, Computer Vision: Algorithms and Applications. Cham, Switzerland: Springer, 2022. 10.1007/978-3-030-34372-9

[6] R. C. Gonzalez and R. E. Woods, Digital Image Processing. Pearson, 2018.

[7] I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning. Cambridge, MA, USA: MIT Press, 2016.

[8] Y. LeCun, Y. Bengio, and G. Hinton, “Deep learning,” Nature, vol. 521, no. 7553, pp. 436–444, May 28 2015. 10.1038/nature14539

[9] R. Girshick, “Rich feature hierarchies for accurate object detection,” in Proc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR), 2014.

[10] J. Redmon and A. Farhadi, “YOLOv3: An incremental improvement,” arXiv preprint, 2018.

[11] A. Bochkovskiy, C.-Y. Wang, and H.-Y. M. Liao, “YOLOv4: Optimal speed and accuracy of object detection,” arXiv preprint, 2020.

[12] R. Ranftl, A. Bochkovskiy, and V. Koltun, “MiDaS: Robust monocular depth estimation,” in Proc. IEEE/CVF Conf. Computer Vision and Pattern Recognition (CVPR), 2020.

[13] A. Grama, A. Gupta, G. Karypis, and V. Kumar, Introduction to Parallel Computing. Pearson Education, 2003.

[14] M. J. Quinn, Parallel Programming in C with MPI and OpenMP. New York, NY, USA: McGraw-Hill, 2004.

[15] A. S. Tanenbaum and H. Bos, Modern Operating Systems. Pearson, 2015.

[16] J. L. Hennessy and D. A. Patterson, Computer Architecture: A Quantitative Approach. Cambridge, MA, USA: Morgan Kaufmann, 2019.

[17] G. M. Amdahl, “Validity of the single processor approach to achieving large scale computing capabilities,” in Proc. AFIPS Spring Joint Computer Conf., 1967. 10.1145/1465482.1465560

[18] D. B. Kirk and W. W. Hwu, Programming Massively Parallel Processors. Cambridge, MA, USA: Morgan Kaufmann, 2016.

[19] Python Software Foundation, “Python documentation: Concurrent execution and threading.” [Online]. Available: https://docs.python.org

[20] Python Software Foundation, “Multiprocessing module documentation.” [Online]. Available: https://docs.python.org/3/library/multiprocessing.html

[21] Y. Zhuang, “Intelligent obstacle avoidance method for blind travel based on an improved YOLO algorithm and binocular vision,” Promet – Traffic&Transportation, vol. 38, no. 5, pp. 1192-1205, 2026. 10.7307/ptt.v38i5.1295

[22] P. Kumari and R. Hammady, “Assisting blind people with AI and audio using smart glasses: System design with YOLOv8 variants comparisons,” Multimedia Syst., vol. 32, no. 1, p. 73, 2026., doi:

[23] Y. Ouyang, R. Zhang, Y. Liu, and Y. Peng, “Research on indoor visual aid algorithms for visually impaired people,” Laser Technology, vol. 49, no. 2, pp. 166–174, 2025.