: pick & place project를 수행하던 도중 6D Pose Estimation에 관심이 갔고, 이에 대해 대표적인 행사가 BOP Challenge라 보기로 하였다.
BOP Challenge: the third in a series of public competitions organized with the goal to capture the status quo in the field of 6D object pose estimation from an RGB-D image.
trend: Methods based on deep neural networks have finally caught up with methods based on point pair features, which were dominating previous editions of the challenge
Point Pair Features: Model Globally, Match Locally: Efficient and Robust 3D Object Recognition paper
match pairs of oriented 3D points btw the point cloud of the test scene and the 3D object model, and aggregate the matches via a voting scheme
DNN-based method
trend: have been commonly trained on "render & paste" images synthesized by openGL rendering of 3D object models randomly positioned on top of random backgrounds
problem: reduce the gap btw the synthetic and real domains
CosyPose: Consistent multi-view multi-object 6D pose estimation paper
DenseFusion: 6D Object Pose Estimation by Iterative Dense Fusion paper
Pix2Pose: Pixel-Wise Coordinate Regression of Objects for 6D Pose Estimation paper
BOP Challenge 2020 on 6D Object Localization
: pick & place project를 수행하던 도중 6D Pose Estimation에 관심이 갔고, 이에 대해 대표적인 행사가 BOP Challenge라 보기로 하였다. BOP Challenge: the third in a series of public competitions organized with the goal to capture the status quo in the field of 6D object pose estimation from an RGB-D image.
trend: Methods based on deep neural networks have finally caught up with methods based on point pair features, which were dominating previous editions of the challenge
DNN-based method
CosyPose: Consistent multi-view multi-object 6D pose estimation paper
DenseFusion: 6D Object Pose Estimation by Iterative Dense Fusion paper
Pix2Pose: Pixel-Wise Coordinate Regression of Objects for 6D Pose Estimation paper