How about the linear evaluation on ImageNet

[Lyan-ing]

How about the linear evaluation on ImageNet? ^_^

[Lyan-ing]

and how to sample the positives in paper section 4.4? thank you

[olivierhenaff]

From section 4.4: "To preserve fine-grained information we sample local features from within each mask and optimize them using the DetCon objective." This includes positive and negative samples.

We didn't evaluate linear classification on ImageNet as we were interested in transfer learning.

[Lyan-ing]

Got it. thank u.