Investigation of the Impact of Synthetic Training Data in the Industrial Application of Terminal Strip Object Detection

arXiv:2403.04809v1 Announce Type: cross
Abstract: In industrial manufacturing, numerous tasks of visually inspecting or detecting specific objects exist that are currently performed manually or by classical image processing methods. Therefore, introducing recent deep learning models to industrial environments holds the potential to increase productivity and enable new applications. However, gathering and labeling sufficient data is often intractable, complicating the implementation of such projects. Hence, image synthesis methods are commonly used to generate synthetic training data from 3D models and annotate …

abstract application arxiv cs.cv cs.lg data deep learning detection environments image image processing impact industrial industrial manufacturing investigation manufacturing object objects processing synthetic tasks terminal training training data type