search

ltttpku / CMMP

16 stars 1 forks source link

data #4

Closed Maritime-Moon closed 1 month ago

Maritime-Moon commented 1 month ago

I should download that VCOCO dataset. Why does the training set in the dataset I downloaded have 82,783 images, and I see that what you are saying seems to be only 10,346 images?

ltttpku commented 1 month ago

The dataset you mentioned is likely MS COCO, while V-COCO is a subset containing 10,346 annotated images. Please refer to their paper or the brief introduction here for more details.