Robust Object Detection and Tracking Model for Visually Impaired People Using Deep Convolution Neural Network Model

Visual Impairment (VI) has become the most significant challenge over the past few years. Even to complete their day-to-day activities, visually impaired persons seek assistance. A currency recognition system and automated common object could enhance the safe movement and transaction people activity with a disability. A model is generated for an object detector that could identify items for people with a disability and other significant uses by identifying them at a certain distance. The present object detection techniques need a massive quantity of training dataset that take longer time, are difficult, and it is a more complicated process. Therefore, in this work, a Robust Object Detection and Tracking for Visually Impaired People using Deep Convolution Neural Network (RODT-DCNN) model is developed. The main purpose of the RODT-DCNN model lies in the accurate identification and classification of objects for VIP. The RODT-DCNN technique comprises two stages of operations such as object detection and object classification. For object detection, YOLO-v5 approach was used. Next, Elman neural network (ENN) approach can be exploited for the effectual identification and classification of objects. At last, sine cosine algorithm (SCA) can be employed for adjusting the parameters of the ENN model. The performance analysis of the RODT-DCNN approach occurs on standard dataset and the outcomes demonstrated the improved detection results of the RODT-DCNN approach in terms of diverse measures.