All reviewers are in unanimous agreement for acceptance. The paper has a number of interesting contributions, mostly empirical, in utilizing a conditional weights network and feature pyramids. As promised in your rebuttal, please release code before acceptance.