At the 6th International ICCV Workshop about "Recovering 6D Pose" MVTec took part in the BOP challenge and won again.
The BOP challenge is a scientific competition where objects have to be found in different data sets. The criteria are the performance and the duration of the algorithm.
Last year we won the contest with our surface-based matching technology. This year we won with a combination of Deep Learning and the classical surface-based matching in the category "fast methods". Our method takes about 0.6 seconds per image, while the best method takes 13.7 seconds per image.
Congratulations to our research department! Learn more about their work here.