We have developed software that recognizes shapes represented by 3D CAD models
in complex real-world tabletop scenes captured by a 3D camera and accurately estimates
their pose with respect to the camera. The software is capable of recognizing multiple
shapes as well as multiple instances of the same shape and it takes only a few seconds to
do so.
****
Cupec, Robert; Vidović, Ivan; Filko, Damir; Đurović, Petra
Object recognition based on convex hull alignment // Pattern recognition, 102 (2020), 107199,
19 doi:10.1016/j.patcog.2020.107199
https://www.sciencedirect.com/science/article/abs/pii/S0031320320300066?via%3Dihub
****