In-house Object Detection System for Visually Impaired

  • Nilakshi Mule, Dipti D. Patil, Yashpalsing D. Chavhan


The worldwide population of visually impaired is 2.2 billion. The visually impaired person has to face many challenges while performing their everyday activities. In this paper, the proposed system mainly focused on providing in-house object detection. The various household objects like TV, Chair, Remote, and Bottle are used as objects for object detection using Raspberry pi 3 kits, Tensorflow, OpenCV, and SSDlite MobileNet V2. The system generates an audio output based on hand gestures and the detected object labeled in English or Hindi languages. The system also calculates the distance between the users and objects. The system has shown precision 0.85 and recalls 0.8 with a 2-second delay in generating audio output.