An assistive model for visually impaired people using YOLO and MTCNN

Visually impaired people face difficulties in safe and independent movement which deprive them from regular professional and social activities in both indoors and outdoors. Similarly they have distressin identification of surrounding environment fundamentals. This paper presents a model to detect brightness and major colors in real-time image by using RGB method by means of an external camera and then identification of fundamental objects as well as facial recognition from personal dataset. For the Object identification and Facial Recognition, YOLO Algorithm and MTCNN Networking are used, respectively. The software support is achieved by using OpenCV libraries of Python as well as implementing machine learning process. The major processor used for our model, Raspberry Pi scans and detects the facial edges via Pi camera and objects in the image are captured and recognized using mobile camera. Image recognition results are transferred to the blind users by means of text-to-speech library. The device portability is achieved by using a battery. The object detection process achieved 6-7 FPS processing with an accuracy rate of 63-80%. The face identification process achieved 80-100% accuracy.