Object Detection and Recognition for Visually Impaired Using Cross-Stage-Partial Network Algorithms

Object detection is a deep learning technology that tracks items in photographs and moving pictures. Object detection algorithms typically confide in artificial intelligence models to produce precise results. These algorithms attempt to replicate human intelligence in recognizing and identifying objects of interest within photographs or videos. This research paper aims to help visually impaired individuals to live independently. We will be utilizing the YOLOv4 algorithm (You Only Look Once). This algorithm employs a modified version of the highly complex Convolutional Neural Network architecture known as DarkNet, along with OpenCV and Google Text to Speech, to convert images to text and then text to speech. The system will continuously capture the live background by using a camera and detect all the entities present, and after successful recognition, the object name would be read out loud by Google Text To Speech. Through a user-friendly gadget with this sophisticated item identification model, the major goal is to enable visually impaired users to comprehend unfamiliar environments.