xyupeng / ContrastiveCrop

[CVPR 2022 Oral] Crafting Better Contrastive Views for Siamese Representation Learning

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

question

cymdhx opened this issue · comments

请问 这个可以用在目标检测上吗。

我还想请教下怎么使用ContrastiveCrop
我看代码里是这个ContrastiveCrop(alpha=alpha, size=224, scale=(0.2, 1.0)),我想问img和box参数是怎么传进去的呢

Yes. Pretraining a model with ContrastiveCrop could potentially improve downstream detection tasks.
Boxes are stored in the dataset, which are passed to ContrastiveCrop as well as other transformations along with imgs. See here