The development and evaluation of an archival system that combines Deep Learning techniques with XR to process and access geo-based temporal 3D scans.

In our age, which is more than ever dynamically archival, digitalisation of our environment and behaviour is the way we teach computers to understand our reality. 3D scanning is an emerging, widely accessible medium, which produces a signalised representation of space and incorporates parameters link to time. This research is focusing on 3D scanning as an archival medium and specifically in the possibilities of combining different techniques to declutter, organise and communicate with a spatio-temporal archive. It aims to develop a conceptual stage of development, identify the key components that it needs to incorporate, build an initial archival system and finally evaluate the performance of all stages. The methodology is divided into two parts: Back and Front end. Here, the former involves the transformation of data into information, through 3D semantic segmentation with deep learning, while the latter involves the transformation of information into knowledge, in terms of communication, through the development of a mobile XR-based system. The case study is a bedroom, but it could be implemented in other contexts such as a museum, a construction site or even a natural environment.