Automated detection of objects using Computer Vision

Kurzfassung

In this master thesis, a method for automatic object recognition of technical components of a complex technical system is developed. For this purpose, a data pool of fused sensor data with different sensors (RGB and infrared), as well as a synthetic data pool with the help of the unreal engine 5 is generated. The object detection takes playce using the Mask R-CNN algorithm created by Matterport. The results are identified objects/pipes, containing in particular information about the shape and position of the objects.