An Intelligent Assistive Navigation Guide System for Blind and Visually Impaired Individuals Using Deep Learning Techniques

To satisfy the need for more advanced and reliable navigation support systems, a growing number of visually impaired individuals face challenges navigating complex environments where traditional assistive tools such as white cane devices are insufficient. This work presents a real-time intelligent assistive navigation system designed to detect obstacles, estimate distances, and provide audio feedback for safe mobility. The system adopts a hybrid Agile and Behaviour-Driven Development approach and leverages YOLOv8 for real-time object detection. An anti-occlusion mechanism enhances recognition of partially hidden objects, while a Distance Estimation Algorithm and Kalman Filter support tracking and motion prediction. A Text-to-Speech module converts detection results into audio guidance. A dataset of 329,067 images from ZED 2i camera captures and Microsoft COCO dataset was used for training. Experimental results show 97.4% precision, 95.2% recall, and 86.9% mAP@0.5. The system demonstrates strong performance in both indoor and outdoor environments, providing accurate obstacle detection, distance estimation, and real-time voice guidance for improved independence and mobility of visually impaired users.

Assistive Navigation System Visual Impairment Deep Learning YOLOv8 Obstacle Detection Kalman Filter Text-to-Speech Computer Vision Object Tracking Real-Time Systems

This study presented the design and implementation of an intelligent assistive navigation system for blind and visually impaired individuals using deep learning techniques. The system addresses limitations of existing assistive technologies by integrating YOLOv8-based object detection, an anti-occlusion enhancement model, distance estimation algorithms, Kalman filter tracking, and a text-to-speech feedback system. The model was trained on a large-scale dataset combining real-world captures and Microsoft COCO data, achieving strong performance with 97.4% precision, 95.2% recall, and 86.9% mAP@0.5. Experimental evaluation confirmed reliable detection of obstacles in diverse environments and effective real-time audio feedback for navigation. The results demonstrate that the proposed system significantly improves mobility, safety, and independence for visually impaired individuals. Future work will focus on expanding dataset diversity, improving robustness under extreme environmental conditions, and optimizing deployment on lightweight wearable devices for real-world adoption.

[1] Anna, F. (2024). World Braille Day; vision loss predicted to surge by 55% in 2050. Statista. https://www.statista.com/chart/31502/expected-number-of-people-with-vision-loss-globally/

[2] Atitallah, A., Said, Y., Atitallah, A., Albekairi, M., Kaaniche, K., & Boubaker, S. (2024). An effective obstacle detection system using deep learning advantages to aid blind and visually impaired navigation. Ain Shams Engineering Journal, 15, 102387. https://doi.org/10.1016/j.asej.2023.102387

[3] Burhan, A. M., Klahan, B., Cummins, W., Andrés-Guerrero, V., Byrne, M. E., O’Reilly, N. J., Chauhan, A., Fitzhenry, L., & Hughes, H. (2021). Posterior segment ophthalmic drug delivery: Role of muco-adhesion with a special focus on chitosan. Pharmaceutics, 13, 1685. https://doi.org/10.3390/pharmaceutics130101685

[4] Cao, F., Xing, B., Luo, J., Li, D., Qian, Y., Zhang, C., Bai, H., & Zhang, H. (2023). An efficient object detection algorithm based on improved YOLOv5 for high-spatial-resolution remote sensing images. Remote Sensing, 15, 3755. https://doi.org/10.3390/rs15153755

[5] Chen, H., Chen, Z., & Yu, H. (2023). Enhanced YOLOv5: An efficient road object detection method. Sensors, 23, 8355. https://doi.org/10.3390/s23208355

[6] Ebere, C. E. U., Udanor, C. N., & Anoliefo, E. (2024). Exploring the depths of visual understanding: A comprehensive review on real-time object detection techniques. Preprints. https://doi.org/10.20944/preprints202402.0583.v1

[7] He, K., Zhang, X., Ren, S., & Sun, J. (2016). Identity mappings in deep residual networks. ECCV, 630–645.

[8] Jiang, P., Ergu, D., Liu, F., Cai, Y., & Ma, B. (2022). A review of YOLO algorithm developments. Procedia Computer Science, 199, 1066–1073. https://doi.org/10.1016/j.procs.2022.01.135

[9] Oluwaseyi, E., Martins, E., & Abraham, E. (2023). A comparative analysis of YOLOv5 and YOLOv7 object detection algorithms. Journal of Computing and Social Informatics, 2(1), 1–12.

[10] WHO. (2023). Blindness and visual impairment. World Health Organization. https://www.who.int/news-room/fact-sheets/detail/blindness-and-visual-impairment

[11] Yu, J., Xu, Z., He, X., Wang, J., Liu, B., Feng, R., Zhu, S., Wang, W., & Li, J. (2023). DIA-TTS: Deep-inherited attention-based text-to-speech synthesizer. Entropy, 25(1), 41. https://doi.org/10.3390/e25010041