CHENG, KAI YANG
National Taipei University of Education Graduate School of Computer Science Student
Email：g110632008@grad.ntue.edu.tw
LIU, YUAN CHEN
National Taipei University of Education Graduate School of Computer Science Professor
Email：liu@tea.ntue.edu.tw
Abstract
For helping people who is blind to face unknown environment with hearing and touching merely, new technology can support not only fast but also convenient service. This research combines with AI and tensorflow in a system of image recognition in a device to serve those blind people, this device is wearable, which is carried on a body. When the person presses the button of device, it will take a photo which represent the person’s vision. The photo will be taken to Raspberry Pi form USB by processor which is processing deep learning and image distinguish. After this process of deep learning and image distinguish, the device will provide prompt voice to remind the person with information of target’s name, amount as well as opposite location. This result of research focus on special object in bathroom. For example, toilets, skins and squat toilets. This data which be collected and the model of deep learning which train with itself are the core in this system. The device goes through from taking photo to supplying voice prompt is just in one second. This device can make invisible people have a better life with image vision tech.
Keywords :Deep Learning, Tensorflow, Blind, Object Detection, Raspberry Pi