Smart Assistive Navigator for the Blind using Image Processing

The life of an individual depends on the basic five senses, and the ability to see is probably the most important one. Visually impaired individuals lack a sense of vision. Hence, daily activities are hampered by their inability to perceive their surroundings. They often face constraints related to independent mobility and navigation. This can lead to difficulties that can only be temporarily subdued by some assisting personnel, creating a feeling of dependency in such an individual. The current methods used nowadays, such as a smart cane, have a limited reach and can only detect obstacles in close proximity to the user. However, to provide comprehensive assistance and enhance independence for visually impaired individuals, it is clear that more techniques and solutions need to be invented. This paper proposes a sophisticated IoT-enabled wearable solution that also encompasses deep learning mechanisms. The architectural design includes a camera module and the Raspberry Pi as the processing unit. It is also equipped with buzzers. The camera module is used to detect obstacles in the user’s route in real-time. Object detection is achieved using OpenCV and deep learning. The use of machine learning here enhances the effectiveness of the application model. Buzzers are interfaced to warn the user of any incoming obstacles. In this way, the system can guide and help the user reach his destination safely and independently, without any external aid.