{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:2IGJVX5IJYG5NPPWLADO7LXE4V","short_pith_number":"pith:2IGJVX5I","canonical_record":{"source":{"id":"2605.05405","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-06T19:43:44Z","cross_cats_sorted":[],"title_canon_sha256":"36ff590c079ad279dfc6b7a807c79e339db6e581406db38d9e9a142b1a66553a","abstract_canon_sha256":"a2137c385d46df452111689804b29aaa46d1b9b316fbc6f4890ab25e1cf649f8"},"schema_version":"1.0"},"canonical_sha256":"d20c9adfa84e0dd6bdf65806efaee4e5430253d2ca2a39be4806efda53ae69e8","source":{"kind":"arxiv","id":"2605.05405","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.05405","created_at":"2026-05-21T01:05:20Z"},{"alias_kind":"arxiv_version","alias_value":"2605.05405v3","created_at":"2026-05-21T01:05:20Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.05405","created_at":"2026-05-21T01:05:20Z"},{"alias_kind":"pith_short_12","alias_value":"2IGJVX5IJYG5","created_at":"2026-05-21T01:05:20Z"},{"alias_kind":"pith_short_16","alias_value":"2IGJVX5IJYG5NPPW","created_at":"2026-05-21T01:05:20Z"},{"alias_kind":"pith_short_8","alias_value":"2IGJVX5I","created_at":"2026-05-21T01:05:20Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:2IGJVX5IJYG5NPPWLADO7LXE4V","target":"record","payload":{"canonical_record":{"source":{"id":"2605.05405","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-06T19:43:44Z","cross_cats_sorted":[],"title_canon_sha256":"36ff590c079ad279dfc6b7a807c79e339db6e581406db38d9e9a142b1a66553a","abstract_canon_sha256":"a2137c385d46df452111689804b29aaa46d1b9b316fbc6f4890ab25e1cf649f8"},"schema_version":"1.0"},"canonical_sha256":"d20c9adfa84e0dd6bdf65806efaee4e5430253d2ca2a39be4806efda53ae69e8","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-21T01:05:20.157959Z","signature_b64":"7J2rfFHj7qyayLEad2fyXAvrolZRxqAcgsESNKBR/hxW48gwGnf0DxwFF/fDZIE3nmF0XqUa/e4NkZKX2MieCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d20c9adfa84e0dd6bdf65806efaee4e5430253d2ca2a39be4806efda53ae69e8","last_reissued_at":"2026-05-21T01:05:20.157213Z","signature_status":"signed_v1","first_computed_at":"2026-05-21T01:05:20.157213Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.05405","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-21T01:05:20Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"psCcA10BDNGRtCkxHL5KQv1/bcurXVVGE61bWgrzs4hyOJJg++E5+Yxjwgkn3L8W7MAxHj+FWl2olte+mVugCg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-29T16:19:41.452198Z"},"content_sha256":"ca69fbeaf716a6b28698571f9c09816ab4c161f0d3f83b67431967694faefbcc","schema_version":"1.0","event_id":"sha256:ca69fbeaf716a6b28698571f9c09816ab4c161f0d3f83b67431967694faefbcc"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:2IGJVX5IJYG5NPPWLADO7LXE4V","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Zero-Shot Satellite Image Retrieval through Joint Embeddings: Application to Crisis Response","license":"http://creativecommons.org/licenses/by/4.0/","headline":"GeoQuery retrieves satellite images matching natural language queries on disasters by aligning text embeddings from a small proxy set with frozen visual embeddings.","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Grace Colverd, James Walsh, Ra\\'ul Ramos-Poll\\'an, William Fawcett","submitted_at":"2026-05-06T19:43:44Z","abstract_excerpt":"Semantic search of Earth observation archives remains challenging. Visual foundation models such as CLAY produce rich embeddings of satellite imagery but lack the natural-language grounding needed for intuitive query, and full contrastive training of a remote-sensing CLIP-style model requires paired data and compute that are unavailable at global scale. To allow natural language querying at global scales, we present GeoQuery, a zero-shot retrieval system that sidesteps data and compute constraints through a two-stage semantic and visual search, leveraging a natural language embedding of a subs"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"On 76 disaster-location queries covering UK floods, US wildfires, and US droughts, GeoQuery achieves 31.6% accuracy within 50 km, with the strongest performance on floods (50% within 50 km).","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That optimizing the description-generation prompt on a 100k proxy subset produces text-embedding distances that correlate sufficiently with frozen CLAY visual distances to support effective two-stage retrieval over the full global archive.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"GeoQuery enables natural-language retrieval of global Sentinel-2 imagery by optimizing text prompts on a 100k proxy subset so that text embeddings correlate with CLAY visual embeddings, then using two-stage text-then-visual search, achieving 31.6% accuracy within 50 km on disaster queries.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"GeoQuery retrieves satellite images matching natural language queries on disasters by aligning text embeddings from a small proxy set with frozen visual embeddings.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"f79c9a60c01673421459d7a42290abdd12a52348596fe28ca9305d6fef9263ea"},"source":{"id":"2605.05405","kind":"arxiv","version":3},"verdict":{"id":"c37d492d-94b4-4a28-a833-9d260f57da6d","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-11T01:54:57.251452Z","strongest_claim":"On 76 disaster-location queries covering UK floods, US wildfires, and US droughts, GeoQuery achieves 31.6% accuracy within 50 km, with the strongest performance on floods (50% within 50 km).","one_line_summary":"GeoQuery enables natural-language retrieval of global Sentinel-2 imagery by optimizing text prompts on a 100k proxy subset so that text embeddings correlate with CLAY visual embeddings, then using two-stage text-then-visual search, achieving 31.6% accuracy within 50 km on disaster queries.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That optimizing the description-generation prompt on a 100k proxy subset produces text-embedding distances that correlate sufficiently with frozen CLAY visual distances to support effective two-stage retrieval over the full global archive.","pith_extraction_headline":"GeoQuery retrieves satellite images matching natural language queries on disasters by aligning text embeddings from a small proxy set with frozen visual embeddings."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.05405/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-20T09:41:54.255300Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T21:01:19.379217Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T13:36:37.017773Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"77b4a2afe14b66750796e37dbdf76cd50762f2d78cef16404f9c0d24c4fdcb33"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"b845a17d32aadb2d7fc3fbdb3c1b09a5f6c7c7760152a81004b8c5ef84b1337a"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"c37d492d-94b4-4a28-a833-9d260f57da6d"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-21T01:05:20Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"Ksu9iCgiiPbB8nZJ0xDBd3XXjFSx5clzbsFde7PMyZPeUaYaa9GW0fBNkXLdNyL9hZ+rMfrzUEgG4+qikLdTCQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-29T16:19:41.453202Z"},"content_sha256":"04aa95c6fea0a046fd5e5b8e06aff105f78db81c6ca0b19f08952ba1e91b9a22","schema_version":"1.0","event_id":"sha256:04aa95c6fea0a046fd5e5b8e06aff105f78db81c6ca0b19f08952ba1e91b9a22"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/2IGJVX5IJYG5NPPWLADO7LXE4V/bundle.json","state_url":"https://pith.science/pith/2IGJVX5IJYG5NPPWLADO7LXE4V/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/2IGJVX5IJYG5NPPWLADO7LXE4V/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-05-29T16:19:41Z","links":{"resolver":"https://pith.science/pith/2IGJVX5IJYG5NPPWLADO7LXE4V","bundle":"https://pith.science/pith/2IGJVX5IJYG5NPPWLADO7LXE4V/bundle.json","state":"https://pith.science/pith/2IGJVX5IJYG5NPPWLADO7LXE4V/state.json","well_known_bundle":"https://pith.science/.well-known/pith/2IGJVX5IJYG5NPPWLADO7LXE4V/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:2IGJVX5IJYG5NPPWLADO7LXE4V","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"a2137c385d46df452111689804b29aaa46d1b9b316fbc6f4890ab25e1cf649f8","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-06T19:43:44Z","title_canon_sha256":"36ff590c079ad279dfc6b7a807c79e339db6e581406db38d9e9a142b1a66553a"},"schema_version":"1.0","source":{"id":"2605.05405","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.05405","created_at":"2026-05-21T01:05:20Z"},{"alias_kind":"arxiv_version","alias_value":"2605.05405v3","created_at":"2026-05-21T01:05:20Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.05405","created_at":"2026-05-21T01:05:20Z"},{"alias_kind":"pith_short_12","alias_value":"2IGJVX5IJYG5","created_at":"2026-05-21T01:05:20Z"},{"alias_kind":"pith_short_16","alias_value":"2IGJVX5IJYG5NPPW","created_at":"2026-05-21T01:05:20Z"},{"alias_kind":"pith_short_8","alias_value":"2IGJVX5I","created_at":"2026-05-21T01:05:20Z"}],"graph_snapshots":[{"event_id":"sha256:04aa95c6fea0a046fd5e5b8e06aff105f78db81c6ca0b19f08952ba1e91b9a22","target":"graph","created_at":"2026-05-21T01:05:20Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"On 76 disaster-location queries covering UK floods, US wildfires, and US droughts, GeoQuery achieves 31.6% accuracy within 50 km, with the strongest performance on floods (50% within 50 km)."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That optimizing the description-generation prompt on a 100k proxy subset produces text-embedding distances that correlate sufficiently with frozen CLAY visual distances to support effective two-stage retrieval over the full global archive."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"GeoQuery enables natural-language retrieval of global Sentinel-2 imagery by optimizing text prompts on a 100k proxy subset so that text embeddings correlate with CLAY visual embeddings, then using two-stage text-then-visual search, achieving 31.6% accuracy within 50 km on disaster queries."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"GeoQuery retrieves satellite images matching natural language queries on disasters by aligning text embeddings from a small proxy set with frozen visual embeddings."}],"snapshot_sha256":"f79c9a60c01673421459d7a42290abdd12a52348596fe28ca9305d6fef9263ea"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"b845a17d32aadb2d7fc3fbdb3c1b09a5f6c7c7760152a81004b8c5ef84b1337a"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"ai_meta_artifact","ran_at":"2026-05-20T09:41:54.255300Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_title_agreement","ran_at":"2026-05-19T21:01:19.379217Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_compliance","ran_at":"2026-05-19T13:36:37.017773Z","status":"completed","version":"1.0.0"}],"endpoint":"/pith/2605.05405/integrity.json","findings":[],"snapshot_sha256":"77b4a2afe14b66750796e37dbdf76cd50762f2d78cef16404f9c0d24c4fdcb33","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Semantic search of Earth observation archives remains challenging. Visual foundation models such as CLAY produce rich embeddings of satellite imagery but lack the natural-language grounding needed for intuitive query, and full contrastive training of a remote-sensing CLIP-style model requires paired data and compute that are unavailable at global scale. To allow natural language querying at global scales, we present GeoQuery, a zero-shot retrieval system that sidesteps data and compute constraints through a two-stage semantic and visual search, leveraging a natural language embedding of a subs","authors_text":"Grace Colverd, James Walsh, Ra\\'ul Ramos-Poll\\'an, William Fawcett","cross_cats":[],"headline":"GeoQuery retrieves satellite images matching natural language queries on disasters by aligning text embeddings from a small proxy set with frozen visual embeddings.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-06T19:43:44Z","title":"Zero-Shot Satellite Image Retrieval through Joint Embeddings: Application to Crisis Response"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.05405","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-11T01:54:57.251452Z","id":"c37d492d-94b4-4a28-a833-9d260f57da6d","model_set":{"reader":"grok-4.3"},"one_line_summary":"GeoQuery enables natural-language retrieval of global Sentinel-2 imagery by optimizing text prompts on a 100k proxy subset so that text embeddings correlate with CLAY visual embeddings, then using two-stage text-then-visual search, achieving 31.6% accuracy within 50 km on disaster queries.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"GeoQuery retrieves satellite images matching natural language queries on disasters by aligning text embeddings from a small proxy set with frozen visual embeddings.","strongest_claim":"On 76 disaster-location queries covering UK floods, US wildfires, and US droughts, GeoQuery achieves 31.6% accuracy within 50 km, with the strongest performance on floods (50% within 50 km).","weakest_assumption":"That optimizing the description-generation prompt on a 100k proxy subset produces text-embedding distances that correlate sufficiently with frozen CLAY visual distances to support effective two-stage retrieval over the full global archive."}},"verdict_id":"c37d492d-94b4-4a28-a833-9d260f57da6d"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:ca69fbeaf716a6b28698571f9c09816ab4c161f0d3f83b67431967694faefbcc","target":"record","created_at":"2026-05-21T01:05:20Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"a2137c385d46df452111689804b29aaa46d1b9b316fbc6f4890ab25e1cf649f8","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-06T19:43:44Z","title_canon_sha256":"36ff590c079ad279dfc6b7a807c79e339db6e581406db38d9e9a142b1a66553a"},"schema_version":"1.0","source":{"id":"2605.05405","kind":"arxiv","version":3}},"canonical_sha256":"d20c9adfa84e0dd6bdf65806efaee4e5430253d2ca2a39be4806efda53ae69e8","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"d20c9adfa84e0dd6bdf65806efaee4e5430253d2ca2a39be4806efda53ae69e8","first_computed_at":"2026-05-21T01:05:20.157213Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-21T01:05:20.157213Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"7J2rfFHj7qyayLEad2fyXAvrolZRxqAcgsESNKBR/hxW48gwGnf0DxwFF/fDZIE3nmF0XqUa/e4NkZKX2MieCQ==","signature_status":"signed_v1","signed_at":"2026-05-21T01:05:20.157959Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.05405","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:ca69fbeaf716a6b28698571f9c09816ab4c161f0d3f83b67431967694faefbcc","sha256:04aa95c6fea0a046fd5e5b8e06aff105f78db81c6ca0b19f08952ba1e91b9a22"],"state_sha256":"defe0750d38a2fd6da4c7cfe15a0c8e6b64a92a1db2ea3633fd2625896fa2832"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"PPxABRsSWQYYVZ4JdzA9HxUZAM1aosWYSN+pZj5CmxWkF/SJM/qHhmDq4iPY/TFrok1eMpXAX5KBbrXdL7EGDA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-29T16:19:41.457463Z","bundle_sha256":"e738957a02d7ada83094f42e53d4d1f68517d43b885e33267dc58dfa8ee6d8cd"}}