Scene understanding and model generation

Abstract ICV scene understanding provides an overview about various object detection and fitting techniques that can produce accurate information for the digital twin (DT) generation of existing assets. The aim of this chapter is to acquire digital twins of existing infrastructure with the least amount of human intervention and give readers the appropriate tools to apply ICV scene understanding and fitting techniques for specific domain applications. The chapter gives an overview of scene understanding for infrastructure applications and user requirements and explores deep learning methods used so far on shape detection and the potential for its use in ICV scene understanding.