Towards Real Time Egocentric Segment Captioning for The Blind and Visually Impaired in RGB-D Theatre Images

In recent years, image captioning and segmentation have emerged as crucial tasks in computer vision, with applications ranging from autonomous driving to content analysis. Although multiple solutions have emerged to help blind and visually impaired people move around their environment, few are applications that help them understand and rebuild a scene in their minds through text. Most built models focus on helping users move and avoid obstacles, restricting the number of environments blind and visually impaired people can be in. In this paper, we will propose an approach that helps them understand their surroundings using image captioning. The particularity of our research is that we offer them descriptions with positions of regions and objects regarding them (left, right, front), as well as positional relationships between regions, while we aim to give them access to theatre plays by applying the solution to our TS-RGBD dataset.

TS-RGBD Dataset: a Novel Dataset for Theatre Scenes Description for People with Visual Impairments

Computer vision was long a tool used for aiding visually impaired people to move around their environment and avoid obstacles and falls. Solutions are limited to either indoor or outdoor scenes, which limits the kind of places and scenes visually disabled people can be in, including entertainment places such as theatres. Furthermore, most of the proposed computer-vision-based methods rely on RGB benchmarks to train their models resulting in a limited performance due to the absence of the depth modality. In this paper, we propose a novel RGB-D dataset containing theatre scenes with ground truth human actions and dense captions annotations for image captioning and human action recognition: TS-RGBD dataset. It includes three types of data: RGB, depth, and skeleton sequences, captured by Microsoft Kinect. We test image captioning models on our dataset as well as some skeleton-based human action recognition models in order to extend the range of environment types where a visually disabled person can be, by detecting human actions and textually describing appearances of regions of interest in theatre scenes.