Hello, thanks for your fantastic work.
When implementing MOCO in your paper, the reference frame and the current frame use different feature extractors. Are their parameter initializations the same? Will the parameters of the two feature extractors be much different after the training? What's more, is there anything you should pay attention to when using MOCO?
Hello, thanks for your fantastic work.
When implementing MOCO in your paper, the reference frame and the current frame use different feature extractors. Are their parameter initializations the same? Will the parameters of the two feature extractors be much different after the training? What's more, is there anything you should pay attention to when using MOCO?
Thank you very much!