TuSimple / centerformer

Implementation for CenterFormer: Center-based Transformer for 3D Object Detection (ECCV 2022)
MIT License
292 stars 28 forks source link

The effect of deformable attention #23

Open star9988rr opened 1 year ago

star9988rr commented 1 year ago

Thank you for your work. I'm a little confused that since the results shown in Tab1 and Tab4 indicate that the deformable attention does not bring benefits, why do you use it?

edwardzhou130 commented 1 year ago

We found deformable attention generally works better in our multi-frame fusion model. We believe this is due to its ability to effectively model cross-attention over a larger range, particularly for fast-moving objects.