Contactless Object Identification Algorithm for the Visually Impaired using EfficientDet

The biggest challenge any Visually Impaired person would face is to perceive the objects around them. The regular way of identifying objects by touching them might not be of favor in every situation. To overcome this problem, we propose a deep learning model for the VI (Visually Impaired) to identify the object, which the person is pointing towards. In this paper, we have developed a contactless solution to detect the objects using transfer learning on the state-of-the-art object detection model, EfficientDet-D7. In addition, we have devised a novel and simple algorithm that can efficiently help in object identification. Simulation results show that the proposed system achieves an mIOU score of 75.85%.