search

alanlukezic / d3s

D3S - Discriminative Single Shot Segmentation Tracker (CVPR 2020)
265 stars 56 forks source link

meaning of rectangles.zip #28

Closed laisimiao closed 3 years ago

laisimiao commented 3 years ago

Hi @alanlukezic Thank you for your work. When training d3s, I don't know the meaning of rectangles.zip you provide us. Taking groundtruth-1.txt for example, does each row means [xmin,ymin,weight,height] of object1 in that sequence? Look forward to your reply!

alanlukezic commented 3 years ago

Since the groundtruth in YT-VOS dataset is given as a segmentation mask and a target bounding box is required for training as well, we provide this information in the groundtruth-*.txt files. Exactly as you said, each row represents an axis-aligned bounding box of the object, which is encoded with 1 in the groundtruth segmentation mask.

laisimiao commented 3 years ago

Ok, thank you!