how to compute object proposals?

"Weakly Supervised Instance Segmentation using Class Peak Response" this paper did not provide the pre-computed object proposals algorithm in their github, and the author also did not answer the issue ZhouYanzhao/PRM#10.
I see you cite Convolutional oriented boundaries and Multiscale combinatorial grouping for image segmentation and object proposal generation, so which one do you use?

