no code implementations • 2 Jun 2022 • Peixian Liang, Yizhe Zhang, Yifan Ding, Jianxu Chen, Chinedu S. Madukoma, Tim Weninger, Joshua D. Shrout, Danny Z. Chen
We observe that probability maps by DL semantic segmentation models can be used to generate many possible instance candidates, and accurate instance segmentation can be achieved by selecting from them a set of "optimized" candidates as output instances.