M2S uses multi-level feature enhancement, auxiliary point cloud reconstruction, and multi-teacher contrastive distillation to boost ego-only 3D mAP by up to 8.64% on V2XSet, V2V4Real, and DAIR-V2X when applied to CoSDH and other detectors.
arXiv preprint arXiv:2203.13168 (2022)
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
-
C2E: Boosting Ego-Only 3D Object Detection via Multi-Teacher Contrastive Knowledge Distillation
M2S uses multi-level feature enhancement, auxiliary point cloud reconstruction, and multi-teacher contrastive distillation to boost ego-only 3D mAP by up to 8.64% on V2XSet, V2V4Real, and DAIR-V2X when applied to CoSDH and other detectors.