FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4386075705> ?p ?o ?g. }

• W4386075705 abstract "Pre-training by numerous image data has become defacto for robust 2D representations. In contrast, due to the expensive data processing, a paucity of 3D datasets severely hinders the learning for high-quality 3D features. In this paper, we propose an alternative to obtain superior 3D representations from 2D pre-trained models via Image-to-Point Masked Autoencoders, named as I2P-MAE. By self-supervised pre-training, we leverage the well learned 2D knowledge to guide 3D masked autoencoding, which reconstructs the masked point tokens with an encoder-decoder architecture. Specifically, we first utilize off-the-shelf 2D models to extract the multi-view visual features of the input point cloud, and then conduct two types of image-to-point learning schemes. For one, we introduce a 2D-guided masking strategy that maintains semantically important point tokens to be visible. Compared to random masking, the network can better concentrate on significant 3D structures with key spatial cues. For another, we enforce these visible tokens to reconstruct multi-view 2D features after the decoder. This enables the network to effectively inherit high-level 2D semantics for discriminative 3D modeling. Aided by our image-to-point pre-training, the frozen I2P-MAE, without any fine-tuning, achieves 93.4% accuracy for linear SVM on ModelNet40, competitive to existing fully trained methods. By further fine-tuning on on ScanObjectNN's hardest split, I2P-MAE attains the state-of-the-art 90.11% accuracy, +3.68% to the second-best, demonstrating superior transferable capacity. Code is available at https://github.com/ZrrSkywalker/I2P-MAE." @default.
• W4386075705 created "2023-08-23" @default.
• W4386075705 creator A5021422000 @default.
• W4386075705 creator A5027948034 @default.
• W4386075705 creator A5029473192 @default.
• W4386075705 creator A5065073978 @default.
• W4386075705 creator A5072861783 @default.
• W4386075705 date "2023-06-01" @default.
• W4386075705 modified "2023-10-01" @default.
• W4386075705 title "Learning 3D Representations from 2D Pre-Trained Models via Image-to-Point Masked Autoencoders" @default.
• W4386075705 cites W1920022804 @default.
• W4386075705 cites W192761727 @default.
• W4386075705 cites W2108598243 @default.
• W4386075705 cites W2117539524 @default.
• W4386075705 cites W2194775991 @default.
• W4386075705 cites W2250384498 @default.
• W4386075705 cites W2412782625 @default.
• W4386075705 cites W2553243490 @default.
• W4386075705 cites W2560722161 @default.
• W4386075705 cites W2796426482 @default.
• W4386075705 cites W2941387379 @default.
• W4386075705 cites W2963150697 @default.
• W4386075705 cites W2963351448 @default.
• W4386075705 cites W2963443993 @default.
• W4386075705 cites W2963509914 @default.
• W4386075705 cites W2963719584 @default.
• W4386075705 cites W2979750740 @default.
• W4386075705 cites W2981440248 @default.
• W4386075705 cites W2982770724 @default.
• W4386075705 cites W2997337685 @default.
• W4386075705 cites W3034459762 @default.
• W4386075705 cites W3035524453 @default.
• W4386075705 cites W3119708198 @default.
• W4386075705 cites W3128716822 @default.
• W4386075705 cites W3153465022 @default.
• W4386075705 cites W3158405343 @default.
• W4386075705 cites W3159481202 @default.
• W4386075705 cites W3171007011 @default.
• W4386075705 cites W3172507977 @default.
• W4386075705 cites W3182683290 @default.
• W4386075705 cites W3197097949 @default.
• W4386075705 cites W3202611145 @default.
• W4386075705 cites W3204568647 @default.
• W4386075705 cites W4200631318 @default.
• W4386075705 cites W4214755140 @default.
• W4386075705 cites W4312317653 @default.
• W4386075705 cites W4313128851 @default.
• W4386075705 cites W4319301012 @default.
• W4386075705 cites W4385768219 @default.
• W4386075705 cites W4386071873 @default.
• W4386075705 doi "https://doi.org/10.1109/cvpr52729.2023.02085" @default.
• W4386075705 hasPublicationYear "2023" @default.
• W4386075705 type Work @default.
• W4386075705 citedByCount "1" @default.
• W4386075705 countsByYear W43860757052023 @default.
• W4386075705 crossrefType "proceedings-article" @default.
• W4386075705 hasAuthorship W4386075705A5021422000 @default.
• W4386075705 hasAuthorship W4386075705A5027948034 @default.
• W4386075705 hasAuthorship W4386075705A5029473192 @default.
• W4386075705 hasAuthorship W4386075705A5065073978 @default.
• W4386075705 hasAuthorship W4386075705A5072861783 @default.
• W4386075705 hasConcept C111919701 @default.
• W4386075705 hasConcept C118505674 @default.
• W4386075705 hasConcept C131979681 @default.
• W4386075705 hasConcept C142362112 @default.
• W4386075705 hasConcept C153083717 @default.
• W4386075705 hasConcept C153180895 @default.
• W4386075705 hasConcept C153349607 @default.
• W4386075705 hasConcept C154945302 @default.
• W4386075705 hasConcept C177264268 @default.
• W4386075705 hasConcept C184337299 @default.
• W4386075705 hasConcept C199360897 @default.
• W4386075705 hasConcept C2524010 @default.
• W4386075705 hasConcept C2776760102 @default.
• W4386075705 hasConcept C2777402240 @default.
• W4386075705 hasConcept C28719098 @default.
• W4386075705 hasConcept C31972630 @default.
• W4386075705 hasConcept C33923547 @default.
• W4386075705 hasConcept C41008148 @default.
• W4386075705 hasConcept C59404180 @default.
• W4386075705 hasConcept C97931131 @default.
• W4386075705 hasConceptScore W4386075705C111919701 @default.
• W4386075705 hasConceptScore W4386075705C118505674 @default.
• W4386075705 hasConceptScore W4386075705C131979681 @default.
• W4386075705 hasConceptScore W4386075705C142362112 @default.
• W4386075705 hasConceptScore W4386075705C153083717 @default.
• W4386075705 hasConceptScore W4386075705C153180895 @default.
• W4386075705 hasConceptScore W4386075705C153349607 @default.
• W4386075705 hasConceptScore W4386075705C154945302 @default.
• W4386075705 hasConceptScore W4386075705C177264268 @default.
• W4386075705 hasConceptScore W4386075705C184337299 @default.
• W4386075705 hasConceptScore W4386075705C199360897 @default.
• W4386075705 hasConceptScore W4386075705C2524010 @default.
• W4386075705 hasConceptScore W4386075705C2776760102 @default.
• W4386075705 hasConceptScore W4386075705C2777402240 @default.
• W4386075705 hasConceptScore W4386075705C28719098 @default.
• W4386075705 hasConceptScore W4386075705C31972630 @default.
• W4386075705 hasConceptScore W4386075705C33923547 @default.
• W4386075705 hasConceptScore W4386075705C41008148 @default.
• W4386075705 hasConceptScore W4386075705C59404180 @default.

• next