Concept for a machine vision framework for production environments based on task-specific synthetic data generation

Procedia CIRP, vol. 138, pp. 1139–1144

Abstract

Already today, Machine Vision (MV) approaches based on Deep Learning (DL) are used increasingly for quality control tasks in production environments. Yet, acquisition and annotation of real-life data necessary for their setup remain extremely challenging due to high costs, required (time) effort, or lack of accessibility to physical systems. This work presents a concept for an MV framework based on task-specific synthetic data generation. Given only available information such as CAD data, an arbitrary amount of annotated data samples can be generated specifically for the considered task, cutting both costs and necessary (time) effort for an industrial MV application setup.

Authors 3

1. Alexander Moriz corresponding Aachen Laboratory for Machine Tools and Production Engineering (WZL)

   RWTH Aachen University

   Affiliation as printed

   Laboratory for Machine Tools and Production Engineering (WZL) of RWTH Aachen University, Campus-Boulevard 30, 52074 Aachen, Germany

2. Dominik Wolfschläger Aachen Laboratory for Machine Tools and Production Engineering (WZL)

   RWTH Aachen University

   Affiliation as printed

   Laboratory for Machine Tools and Production Engineering (WZL) of RWTH Aachen University, Campus-Boulevard 30, 52074 Aachen, Germany

3. Robert Heinrich Schmitt Aachen Frauehofer Institute for Production Metrology IPT Laboratory for Machine Tools and Production Engineering (WZL)

   RWTH Aachen University · Fraunhofer Institute for Production Technology IPT

   Affiliation as printed

   Frauehofer Institute for Production Metrology IPT, Steinbachstrasse 17, 52074 Aachen, Germany

   Laboratory for Machine Tools and Production Engineering (WZL) of RWTH Aachen University, Campus-Boulevard 30, 52074 Aachen, Germany