[Feature Request] Separate Region_caption objects

[toyxyz]

Thanks for the great work as always!

Is it possible to separate the masks of only certain captions - for example, 'person' - in Region_caption?

Currently, the regions of all captions are combined into one mask.