facebookresearch / ContrastiveSceneContexts

Code for CVPR 2021 oral paper "Exploring Data-Efficient 3D Scene Understanding with Contrastive Scene Contexts"
MIT License
224 stars 28 forks source link

Which sampling method does ‘Ours(init)’ use in Figure 6? #24

Closed WeijianZhang123 closed 2 years ago

WeijianZhang123 commented 2 years ago

image

The original text is 'Ours (init) denotes the network initialization by our pre-trained model'.

The x-axis is limited labeled points, so which sampling method does 'Ours(init)' use to get these limited labeled points?

Thanks!

Hiusam commented 2 years ago

I want to ask does Ours(act. labeling) mean train from scratch using data selected by the pre-train network?

Sekunde commented 2 years ago

@Hiusam hi, you understand it correctly.