D. Palogiannidis1

1 St. Petersburg Electrotechnical University "LETI" (St. Petersburg, Russia)
1 dimitris.palogiannidis@gmail.com

Visual impairment affects over 2.2 billion people and limits mobility, autonomy, and mental well-being. Many assistive technologies still depend on user interpretation of feedback, increasing cognitive load. This work presents an automated object avoidance system for visually impaired users, combining YOLOv8-based computer vision with real-time audio navigation. Trained on over 14,000 annotated images, the system uses spatial sector analysis and a weighted decision tree to generate voice commands. It prioritizes safety by addressing central obstacles and adapting guidance based on object type and confidence. Audio prompts with multilingual support enhance usability. The system achieved 70.56% precision, 93.96% recall, and an F1 score of 80.59%, with 48.7% accuracy in navigation commands and a 14.4% critical misclassification rate. The results highlight its potential for accessible, real-time mobility assistance.

Palogiannidis D. Processing and analysis of video images using artificial intelligence in a spatial navigation system for people with visual impairments. Biomedicine Radioengineering. 2025. V. 28. № 5. P. 167–172. DOI: https:// doi.org/10.18127/ j15604136-202505-33 (In Russian)

1. Narusheniya zreniya i slepota [Elektronnyj resurs]. URL: https://www.who.int/news-room/fact-sheets/detail/blindness-and-visual-impairment (In Russian).
2. Press V.G. i dr. Osobennosti pacientov s plohim zreniem: vysokaya cennost'. Zhurnal gospital'noj mediciny. 2015. T. 10. № 5. S. 311–313 (In Russian).
3. Demmin D.L., Sil'verstajn S.M. Narusheniya zreniya i psihicheskoe zdorov'e: neudovletvorennye potrebnosti i varianty lecheniya. Klinicheskaya oftal'mologiya. 2020. T. 14. S. 4229–4251 (In Russian).
4. MakDonnall M.S., MakNajt Chzh.S. Svyaz' mezhdu imeyushchimisya narusheniyami zreniya, zdorov'em i statusom zanyatosti. 2021 [Elektronnyj resurs]. URL: https://journals.sagepub.com/doi/10.1177/0145482X211016570 (In Russian).
5. Kim H., Ku H., Han E. Social'no-ekonomicheskie i fizicheskie izmeneniya sostoyaniya zdorov'ya posle narusheniya zreniya v Koree s ispol'zovaniem metoda raznosti-raznostej. Scientific Reports. 2021. T. 11. № 1. S. 820 (In Russian).
6. Hsie I.-H. i dr. Nosimaya vspomogatel'naya sistema na osnove svertochnyh nejronnyh setej dlya slabovidyashchih lyudej pri peredvizhenii na ulice. Applied Sciences. 2021. T. 11. № 21 (In Russian).
7. Chang V.-Zh. i dr. Vspomogatel'naya sistema na osnove iskusstvennogo intellekta i edge-komp'yutinga dlya obespecheniya bezopasnosti slabovidyashchih peshekhodov na peshekhodnyh perekhodah. IEEE Transactions on Consumer Electronics. 2021. T. 67, № 1.
   S. 3–11 (In Russian).
8. Sistema prostranstvennoj osvedomlennosti dlya slabovidyashchih dlya navigacii v pomeshcheniyah i povsednevnoj deyatel'nosti [Elektronnyj resurs]. URL: https://www.mdpi.com/2313-433X/11/1/9 (In Russian).
9. Olujele S., Adejandzhu I., Sobovale A. Robotizirovannyj pomoshchnik dlya raspoznavaniya ob"ektov s ispol'zovaniem svertochnoj nejronnoj seti: 1. ABUAD Journal of Engineering Research and Development (AJERD). 2024. T. 7. № 1. S. 1–13 (In Russian).
10. Dzhocher G., Cyu Dzh., Chaurasiya A. Ultralytics YOLO: Python. 2023.
11. Dannye izobrazhenij (obnaruzhenie ob"ektov i sozdanie podpisej) [Elektronnyj resurs]. URL: https://www.kaggle.com/datasets/ aruneshhh/object-detection-images (In Russian).
12. TensorFlow [Elektronnyj resurs]. URL: https://www.tensorflow.org/.
13. Lokalizaciya – dokumentaciya Pygame GUI 0.6.14 [Elektronnyj resurs]. URL: https://pygame-gui.readthedocs.io/en/latest/localization.html (In Russian).