Vision-Based Assistive Systems for Visually Impaired People: A Review

This research reviews the current state of vision-based assistive solutions for the visually impaired (VI). The paper focuses primarily on camera-based assistive system solutions. We focused the review on vision-based assistive solutions proposed for VI people. The sensors, image processing algorithms, and wireless communication protocols employed in the survey have been summarised. Acoustic output devices were used in addition to cameras, Radio Frequency Identification (RFID), and Global Positioning System (GPS). Vision-based assistive solutions have evolved from traditional image processing techniques to machine learning to deep learning for assistance for VI users. Wi-Fi and Bluetooth devices are the most common wireless technologies used by vision-based assistive systems. The literature does not adequately leverage the optimization of deep learning models for edge devices.