Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386076357> ?p ?o ?g. }
Showing items 1 to 96 of
96
with 100 items per page.
- W4386076357 abstract "This paper introduces the Masked Voxel Jigsaw and Reconstruction (MV-JAR) method for LiDAR-based self-supervised pre-training and a carefully designed data-efficient 3D object detection benchmark on the Waymo dataset. Inspired by the scene-voxel-point hierarchy in downstream 3D object detectors, we design masking and reconstruction strategies accounting for voxel distributions in the scene and local point distributions within the voxel. We employ a Reversed-Furthest-Voxel-Sampling strategy to address the uneven distribution of LiDAR points and propose MV-JAR, which combines two techniques for modeling the aforementioned distributions, resulting in superior performance. Our experiments reveal limitations in previous data-efficient experiments, which uniformly sample fine-tuning splits with varying data proportions from each LiDAR sequence, leading to similar data diversity across splits. To address this, we propose a new benchmark that samples scene sequences for diverse fine-tuning splits, ensuring adequate model convergence and providing a more accurate evaluation of pre-training methods. Experiments on our Waymo benchmark and the KITTI dataset demonstrate that MV-JAR consistently and significantly improves 3D detection performance across various data scales, achieving up to a 6.3% increase in mAPH compared to training from scratch. Codes and the benchmark are available at https://github.com/SmartBot-PJLab/MV-JAR." @default.
- W4386076357 created "2023-08-23" @default.
- W4386076357 creator A5000349453 @default.
- W4386076357 creator A5010087030 @default.
- W4386076357 creator A5020672335 @default.
- W4386076357 creator A5053876501 @default.
- W4386076357 creator A5064647266 @default.
- W4386076357 creator A5068413732 @default.
- W4386076357 creator A5075229879 @default.
- W4386076357 date "2023-06-01" @default.
- W4386076357 modified "2023-10-02" @default.
- W4386076357 title "MV-JAR: Masked Voxel Jigsaw and Reconstruction for LiDAR-Based Self-Supervised Pre-Training" @default.
- W4386076357 cites W2150066425 @default.
- W4386076357 cites W2796426482 @default.
- W4386076357 cites W2897529137 @default.
- W4386076357 cites W2963727135 @default.
- W4386076357 cites W2968296999 @default.
- W4386076357 cites W3035172746 @default.
- W4386076357 cites W3035524453 @default.
- W4386076357 cites W3119708198 @default.
- W4386076357 cites W3138516171 @default.
- W4386076357 cites W3166470370 @default.
- W4386076357 cites W3166573884 @default.
- W4386076357 cites W3167095230 @default.
- W4386076357 cites W3197097949 @default.
- W4386076357 cites W3203240897 @default.
- W4386076357 cites W4214624153 @default.
- W4386076357 cites W4214777292 @default.
- W4386076357 cites W4312270234 @default.
- W4386076357 cites W4312307873 @default.
- W4386076357 cites W4312804044 @default.
- W4386076357 cites W4312972416 @default.
- W4386076357 cites W4313156423 @default.
- W4386076357 doi "https://doi.org/10.1109/cvpr52729.2023.01292" @default.
- W4386076357 hasPublicationYear "2023" @default.
- W4386076357 type Work @default.
- W4386076357 citedByCount "0" @default.
- W4386076357 crossrefType "proceedings-article" @default.
- W4386076357 hasAuthorship W4386076357A5000349453 @default.
- W4386076357 hasAuthorship W4386076357A5010087030 @default.
- W4386076357 hasAuthorship W4386076357A5020672335 @default.
- W4386076357 hasAuthorship W4386076357A5053876501 @default.
- W4386076357 hasAuthorship W4386076357A5064647266 @default.
- W4386076357 hasAuthorship W4386076357A5068413732 @default.
- W4386076357 hasAuthorship W4386076357A5075229879 @default.
- W4386076357 hasConcept C106131492 @default.
- W4386076357 hasConcept C127313418 @default.
- W4386076357 hasConcept C131979681 @default.
- W4386076357 hasConcept C13280743 @default.
- W4386076357 hasConcept C140779682 @default.
- W4386076357 hasConcept C145420912 @default.
- W4386076357 hasConcept C153180895 @default.
- W4386076357 hasConcept C154945302 @default.
- W4386076357 hasConcept C185798385 @default.
- W4386076357 hasConcept C205649164 @default.
- W4386076357 hasConcept C2779405079 @default.
- W4386076357 hasConcept C31972630 @default.
- W4386076357 hasConcept C33923547 @default.
- W4386076357 hasConcept C41008148 @default.
- W4386076357 hasConcept C51399673 @default.
- W4386076357 hasConcept C54170458 @default.
- W4386076357 hasConcept C62649853 @default.
- W4386076357 hasConceptScore W4386076357C106131492 @default.
- W4386076357 hasConceptScore W4386076357C127313418 @default.
- W4386076357 hasConceptScore W4386076357C131979681 @default.
- W4386076357 hasConceptScore W4386076357C13280743 @default.
- W4386076357 hasConceptScore W4386076357C140779682 @default.
- W4386076357 hasConceptScore W4386076357C145420912 @default.
- W4386076357 hasConceptScore W4386076357C153180895 @default.
- W4386076357 hasConceptScore W4386076357C154945302 @default.
- W4386076357 hasConceptScore W4386076357C185798385 @default.
- W4386076357 hasConceptScore W4386076357C205649164 @default.
- W4386076357 hasConceptScore W4386076357C2779405079 @default.
- W4386076357 hasConceptScore W4386076357C31972630 @default.
- W4386076357 hasConceptScore W4386076357C33923547 @default.
- W4386076357 hasConceptScore W4386076357C41008148 @default.
- W4386076357 hasConceptScore W4386076357C51399673 @default.
- W4386076357 hasConceptScore W4386076357C54170458 @default.
- W4386076357 hasConceptScore W4386076357C62649853 @default.
- W4386076357 hasFunder F4320321920 @default.
- W4386076357 hasLocation W43860763571 @default.
- W4386076357 hasOpenAccess W4386076357 @default.
- W4386076357 hasPrimaryLocation W43860763571 @default.
- W4386076357 hasRelatedWork W2058517172 @default.
- W4386076357 hasRelatedWork W2160739731 @default.
- W4386076357 hasRelatedWork W2979718872 @default.
- W4386076357 hasRelatedWork W3127007190 @default.
- W4386076357 hasRelatedWork W3158534694 @default.
- W4386076357 hasRelatedWork W3206828132 @default.
- W4386076357 hasRelatedWork W4287365848 @default.
- W4386076357 hasRelatedWork W4290774832 @default.
- W4386076357 hasRelatedWork W4309089738 @default.
- W4386076357 hasRelatedWork W4312559648 @default.
- W4386076357 isParatext "false" @default.
- W4386076357 isRetracted "false" @default.
- W4386076357 workType "article" @default.