Matches in SemOpenAlex for { <https://semopenalex.org/work/W4281768077> ?p ?o ?g. }
Showing items 1 to 72 of
72
with 100 items per page.
- W4281768077 abstract "This article addresses the problem of distilling knowledge from a large teacher model to a slim student network for LiDAR semantic segmentation. Directly employing previous distillation approaches yields inferior results due to the intrinsic challenges of point cloud, i.e., sparsity, randomness and varying density. To tackle the aforementioned problems, we propose the Point-to-Voxel Knowledge Distillation (PVD), which transfers the hidden knowledge from both point level and voxel level. Specifically, we first leverage both the pointwise and voxelwise output distillation to complement the sparse supervision signals. Then, to better exploit the structural information, we divide the whole point cloud into several supervoxels and design a difficulty-aware sampling strategy to more frequently sample supervoxels containing less-frequent classes and faraway objects. On these supervoxels, we propose inter-point and inter-voxel affinity distillation, where the similarity information between points and voxels can help the student model better capture the structural information of the surrounding environment. We conduct extensive experiments on two popular LiDAR segmentation benchmarks, i.e., nuScenes and SemanticKITTI. On both benchmarks, our PVD consistently outperforms previous distillation approaches by a large margin on three representative backbones, i.e., Cylinder3D, SPVNAS and MinkowskiNet. Notably, on the challenging nuScenes and SemanticKITTI datasets, our method can achieve roughly 75% MACs reduction and 2x speedup on the competitive Cylinder3D model and rank 1st on the SemanticKITTI leaderboard among all published algorithms. Our code is available at https://github.com/cardwing/Codes-for-PVKD." @default.
- W4281768077 created "2022-06-13" @default.
- W4281768077 creator A5005226727 @default.
- W4281768077 creator A5005264558 @default.
- W4281768077 creator A5005626854 @default.
- W4281768077 creator A5075925819 @default.
- W4281768077 creator A5091374776 @default.
- W4281768077 date "2022-06-05" @default.
- W4281768077 modified "2023-09-26" @default.
- W4281768077 title "Point-to-Voxel Knowledge Distillation for LiDAR Semantic Segmentation" @default.
- W4281768077 doi "https://doi.org/10.48550/arxiv.2206.02099" @default.
- W4281768077 hasPublicationYear "2022" @default.
- W4281768077 type Work @default.
- W4281768077 citedByCount "0" @default.
- W4281768077 crossrefType "posted-content" @default.
- W4281768077 hasAuthorship W4281768077A5005226727 @default.
- W4281768077 hasAuthorship W4281768077A5005264558 @default.
- W4281768077 hasAuthorship W4281768077A5005626854 @default.
- W4281768077 hasAuthorship W4281768077A5075925819 @default.
- W4281768077 hasAuthorship W4281768077A5091374776 @default.
- W4281768077 hasBestOaLocation W42817680771 @default.
- W4281768077 hasConcept C119857082 @default.
- W4281768077 hasConcept C127313418 @default.
- W4281768077 hasConcept C131979681 @default.
- W4281768077 hasConcept C153083717 @default.
- W4281768077 hasConcept C153180895 @default.
- W4281768077 hasConcept C154945302 @default.
- W4281768077 hasConcept C165696696 @default.
- W4281768077 hasConcept C178790620 @default.
- W4281768077 hasConcept C185592680 @default.
- W4281768077 hasConcept C204030448 @default.
- W4281768077 hasConcept C31972630 @default.
- W4281768077 hasConcept C38652104 @default.
- W4281768077 hasConcept C41008148 @default.
- W4281768077 hasConcept C51399673 @default.
- W4281768077 hasConcept C54170458 @default.
- W4281768077 hasConcept C62649853 @default.
- W4281768077 hasConcept C89600930 @default.
- W4281768077 hasConceptScore W4281768077C119857082 @default.
- W4281768077 hasConceptScore W4281768077C127313418 @default.
- W4281768077 hasConceptScore W4281768077C131979681 @default.
- W4281768077 hasConceptScore W4281768077C153083717 @default.
- W4281768077 hasConceptScore W4281768077C153180895 @default.
- W4281768077 hasConceptScore W4281768077C154945302 @default.
- W4281768077 hasConceptScore W4281768077C165696696 @default.
- W4281768077 hasConceptScore W4281768077C178790620 @default.
- W4281768077 hasConceptScore W4281768077C185592680 @default.
- W4281768077 hasConceptScore W4281768077C204030448 @default.
- W4281768077 hasConceptScore W4281768077C31972630 @default.
- W4281768077 hasConceptScore W4281768077C38652104 @default.
- W4281768077 hasConceptScore W4281768077C41008148 @default.
- W4281768077 hasConceptScore W4281768077C51399673 @default.
- W4281768077 hasConceptScore W4281768077C54170458 @default.
- W4281768077 hasConceptScore W4281768077C62649853 @default.
- W4281768077 hasConceptScore W4281768077C89600930 @default.
- W4281768077 hasLocation W42817680771 @default.
- W4281768077 hasLocation W42817680772 @default.
- W4281768077 hasOpenAccess W4281768077 @default.
- W4281768077 hasPrimaryLocation W42817680771 @default.
- W4281768077 hasRelatedWork W1669643531 @default.
- W4281768077 hasRelatedWork W2005437358 @default.
- W4281768077 hasRelatedWork W2008656436 @default.
- W4281768077 hasRelatedWork W2517104666 @default.
- W4281768077 hasRelatedWork W2976989770 @default.
- W4281768077 hasRelatedWork W3083637823 @default.
- W4281768077 hasRelatedWork W4226195147 @default.
- W4281768077 hasRelatedWork W4286225597 @default.
- W4281768077 hasRelatedWork W4310264062 @default.
- W4281768077 hasRelatedWork W4312828305 @default.
- W4281768077 isParatext "false" @default.
- W4281768077 isRetracted "false" @default.
- W4281768077 workType "article" @default.