{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2022:2FIYED3BC2IDVQDL6DKTVOJWQQ","short_pith_number":"pith:2FIYED3B","schema_version":"1.0","canonical_sha256":"d151820f6116903ac06bf0d53ab9368419b4a4a1efebf84b6812a0175b8906f5","source":{"kind":"arxiv","id":"2212.06785","version":1},"attestation_state":"computed","paper":{"title":"Learning 3D Representations from 2D Pre-trained Models via Image-to-Point Masked Autoencoders","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Hongsheng Li, Liuhui Wang, Peng Gao, Renrui Zhang, Yu Qiao","submitted_at":"2022-12-13T17:59:20Z","abstract_excerpt":"Pre-training by numerous image data has become de-facto for robust 2D representations. In contrast, due to the expensive data acquisition and annotation, a paucity of large-scale 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 architectu"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2212.06785","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2022-12-13T17:59:20Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"bfa1e1b0a6582b6b11fa7e5b84a5b5830d1c8d85c2a2722844f152909ecbe760","abstract_canon_sha256":"7cd5dc6b2142dd0031ef10ee5d3d05758a66274fd670872da69db5f599ac26a5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:24:54.594347Z","signature_b64":"VrMhOU0EdU6Zt6nHxh92YqNfr63idpVsfhBXPKfe7U6bfp/ztMTjBDGKx5eiNakIDzZPjPV+7CfyeASO9MClDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d151820f6116903ac06bf0d53ab9368419b4a4a1efebf84b6812a0175b8906f5","last_reissued_at":"2026-07-05T05:24:54.593854Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:24:54.593854Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning 3D Representations from 2D Pre-trained Models via Image-to-Point Masked Autoencoders","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Hongsheng Li, Liuhui Wang, Peng Gao, Renrui Zhang, Yu Qiao","submitted_at":"2022-12-13T17:59:20Z","abstract_excerpt":"Pre-training by numerous image data has become de-facto for robust 2D representations. In contrast, due to the expensive data acquisition and annotation, a paucity of large-scale 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 architectu"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2212.06785","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2212.06785/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2212.06785","created_at":"2026-07-05T05:24:54.593914+00:00"},{"alias_kind":"arxiv_version","alias_value":"2212.06785v1","created_at":"2026-07-05T05:24:54.593914+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2212.06785","created_at":"2026-07-05T05:24:54.593914+00:00"},{"alias_kind":"pith_short_12","alias_value":"2FIYED3BC2ID","created_at":"2026-07-05T05:24:54.593914+00:00"},{"alias_kind":"pith_short_16","alias_value":"2FIYED3BC2IDVQDL","created_at":"2026-07-05T05:24:54.593914+00:00"},{"alias_kind":"pith_short_8","alias_value":"2FIYED3B","created_at":"2026-07-05T05:24:54.593914+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2303.16199","citing_title":"LLaMA-Adapter: Efficient Fine-tuning of Language Models with Zero-init Attention","ref_index":190,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/2FIYED3BC2IDVQDL6DKTVOJWQQ","json":"https://pith.science/pith/2FIYED3BC2IDVQDL6DKTVOJWQQ.json","graph_json":"https://pith.science/api/pith-number/2FIYED3BC2IDVQDL6DKTVOJWQQ/graph.json","events_json":"https://pith.science/api/pith-number/2FIYED3BC2IDVQDL6DKTVOJWQQ/events.json","paper":"https://pith.science/paper/2FIYED3B"},"agent_actions":{"view_html":"https://pith.science/pith/2FIYED3BC2IDVQDL6DKTVOJWQQ","download_json":"https://pith.science/pith/2FIYED3BC2IDVQDL6DKTVOJWQQ.json","view_paper":"https://pith.science/paper/2FIYED3B","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2212.06785&json=true","fetch_graph":"https://pith.science/api/pith-number/2FIYED3BC2IDVQDL6DKTVOJWQQ/graph.json","fetch_events":"https://pith.science/api/pith-number/2FIYED3BC2IDVQDL6DKTVOJWQQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/2FIYED3BC2IDVQDL6DKTVOJWQQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/2FIYED3BC2IDVQDL6DKTVOJWQQ/action/storage_attestation","attest_author":"https://pith.science/pith/2FIYED3BC2IDVQDL6DKTVOJWQQ/action/author_attestation","sign_citation":"https://pith.science/pith/2FIYED3BC2IDVQDL6DKTVOJWQQ/action/citation_signature","submit_replication":"https://pith.science/pith/2FIYED3BC2IDVQDL6DKTVOJWQQ/action/replication_record"}},"created_at":"2026-07-05T05:24:54.593914+00:00","updated_at":"2026-07-05T05:24:54.593914+00:00"}