In the just released collected papers of the European Conference on Computer Vision (ECCV2022), the joint academic achievement of the Technical University of Munich, Meta-Bounds and Google was accepted. As one of the world’s top 3 computer vision conferences, this edition of ECCV received 5,803 effective papers, of which 1,650 were accepted, with an acceptance rate of 28% only. Meta-Bounds will keep exploring the technology frontier, and participate actively in academic exchanges and international cooperation.
Exploring the frontier with a view to the future
This paper proposes a graphic relational structure E-Graph, which abstracts structural information in a scene to eliminate co-vision reliance in posture estimation, help the system realize low-drift rotational estimation, and improve the accuracy and efficiency of system 6-DoF posture estimation, effectively solving the problem of gradual drifting of posture estimation results when the SLAM system is unable to loop back for a long time.
Empowering the industry through prospective thinking
Technological innovation is the primary impetus to industry development. Meta-Bounds will input more and more scientific research resources to accelerate the commercialization of cutting-edge achievements beginning with scientific research, join hands with customers to further meet diversified consumer needs for AR glasses, and contribute to technological breakthroughs in the whole industry!