angle-double-right-blueangle-double-rightangle-down-regularangle-right-duotoneangle-right-regularangle-up-regularcontact-card-brandsenvelope-solidfacebook-f-brandsfile-pdf-lightglobe-americas-regulargoogle-plus-g-brandsinstagram-brandskununu-brandslinkedin-in-brandsmap-marker-altplay-circleplus-solidrss-brandssearch-regulartwitter-brandsuser-altuserxing-brandsyoutube-brands

Quality control of finished vehicles through the analysis of vehicle images

Automotive OEM

Camera-based damage detection for quality control in automobiles

Implemented a proof-of-concept in the quality control of finished vehicles by analyzing images of the vehicle:

  • Agile control and project implementation
  • Analyzed and evaluated the first heterogeneous data supply (different exposure, small number of images, different positions, etc.) for suitability in the development of a neural network
  • Extensive data engineering to absorb the limited quality of data: Background isolation, neutralization of different exposures through edge formation, formation of similar clusters with respect to exposure and vehicle type, etc.
  • Successful development of a neural network for dirt and scratch detection in vehicles
  • Recommendations for action and process optimization for a future operational implementation

Further informationen

Your Contacts

To the top

Contact
us