Computer Vision serving visually impaired people

To explore and recognize 3D objects, various solutions exist, such as the use of 3D glasses and immersive glasses. If these devices are suitable for persons without visual impairment, they still unusable for those with visual impairments. In this work, we propose an alternative device adapted to these people and which is based on the use of a 2D screen. In order to make the recognition of 3D objects possible on ordinary 2D screens, the renderings of 3D objects are augmented by image processing and a suitable interface. This tool, developed under Unity, makes it possible to exploit the wealth of information inherent in 3D models without having to resort to segmentation and voice transcription. The contribution of this article lies in the co-design of the tool with visually impaired people, in the proposal of a tool that can be parameterized and customized according to needs, and in its evaluation in an ecological environment by visually impaired people. The results of this evaluation confirm the value of the tool for optimizing the residual vision of visually impaired people.