search

mecarill / 2pcnet

Other
84 stars 22 forks source link

Use of two-stage consistency #7

Open ShenZheng2000 opened 1 year ago

ShenZheng2000 commented 1 year ago

The paper demonstrates the significant influence of two-stage consistency on mAP. However, in the code, the "CONSISTENCY: True" configuration parameter is not utilized.

Moreover, I am curious about the performance of two-stage consistency when compared to one-stage consistency, which is commonly employed by many methods, as opposed to no consistency at all.

image

ShenZheng2000 commented 1 year ago

On reading the paper again, it appears the last row of Table 4 indicates a base mean teacher baseline like [1], I'm assuming this is akin to just employing Phase 1 consistency (or One stage consistency; indicated in Fig 2 of your paper).

However, I don't see a simple way to train this baseline. I'm would like to do that, could you please point to the script and the corresponding changes I might need to make. Thanks!

[1] Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results, NeurIPS 2017