princeton-vl / CornerNet-Lite

BSD 3-Clause "New" or "Revised" License
1.78k stars 431 forks source link

about course bbox prediction #73

Closed Stick-To closed 5 years ago

Stick-To commented 5 years ago

How to get bbox from attention map, does it predict width and height of target object?
And the locations obtrained from the bounding boes prediction, is it a small cornernet or anchor-based net? And is the network for processing croped subimage a cornetnt?