Adding multi-scale attention and upsampling to 3DETR improves mAP@50 by 4.78% on ScanNetv2 for 3D object detection.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
LOD-Net: Locality-Aware 3D Object Detection Using Multi-Scale Transformer Network
Adding multi-scale attention and upsampling to 3DETR improves mAP@50 by 4.78% on ScanNetv2 for 3D object detection.