CVision is a deep learning based Python program designed to assist the visually-impaired in navigating their surroundings. Real-time video from a portable camera is processed to generate directional sound cues to be heard on headphones. These sounds reflect the relative locations and distances of detected objects. Potential dangers such as very close objects are alerted to the user.

CVision offers two modes: a general mode where the user is notified of all visible objects and a tracking mode which guides the user towards the selected object by a sine tone. CVision also supports ArUco markers, unique square patterns, which may be placed in public areas to reliably notify users of important landmarks such as building entrances.

By combining the latest deep learning models, spatial sound, and marker detection, CVision aims to deliver a real-time auditory experience, empowering users to navigate their environments with more confidence and safety.