Audio Feedback Through Realtime Object Detection Using Yolov5
P. Nagamani, V. Anusha, Md. Shabeena, B. Srimivas Raja, T. Niteesh Kumar, G. Durga Prasad · 2024
Abstract-This study offers a novel approach that combines Google Text-to-Speech (GTTS) with YOLOv5 (You Only Look One) object identification to improve accessibility for those with visual impairments. By using object detection and sound recognition to provide real-time information about their surroundings, the suggested system empowers blind individuals. Leveraging its efficiency and speed, the YOLOv5 model is used to reliably identify and categorize items inside the user’s surroundings. To improve situational awareness, sound recognition algorithms are integrated in tandem to recognize and analyse auditory signals, such as alerts, sirens, or other significant noises. Detected items are translated into intelligible spoken descriptions using GTTS to fill in the visual information gap and provide an aural comprehension of the environment. The realtime operation of the system guarantees the prompt and pertinent dissemination of information. Extensive testing with visually impaired participants is used to assess the efficacy of the proposed solution, with a focus on user input and system response. The results show how the system can greatly enhance the quality of life for those who are visually impaired by providing them with a thorough and intuitive grasp of their surroundings.