Add leak_class_ids.json

leak_class_ids.json

@@ -0,0 +1,486 @@
+{
+ "name": "GLDv2-clean classes overlapping the ROxford and RParis benchmarks",
+ "purpose": "The class ids excluded from the decontaminated (decon) training run. Published so the decon protocol can be checked independently rather than taken on trust.",
+ "totals": {
+  "leak_classes": 87,
+  "leak_images": 2529,
+  "gldv2_clean_classes": 81313,
+  "gldv2_clean_images": 1580470
+ },
+ "comparison": {
+  "rgldv2_clean": {
+   "images_removed": 1565,
+   "landmark_ids_removed": 18,
+   "source": "Song et al., CVPR 2024, page 4"
+  },
+  "note": "Our match set is NOT a superset of RGLDv2-clean: it contains 10 of their 18 landmark ids and 1,216 of their 1,565 images. Image-level verification puts this matcher at 43% precision and 76% recall over contamination, so the classes below define a conservative training ablation rather than a stricter cleanup. See the model card's Evaluation integrity section and fusion-data:/fp_phase3a/leak_verify/leak_verify_results.json."
+ },
+ "method": {
+  "source_field": "GLDv2 landmark categories (Wikimedia Commons category URLs) matched against the landmarks the two benchmarks query.",
+  "rules": {
+   "blanket:<city>": "category is an Oxford landmark and ROxford queries Oxford landmarks generally, so the whole class is excluded",
+   "unique:<landmark>": "category names a specific landmark that appears in the benchmark query set",
+   "ctx:<city>+<landmark>": "category names both the city and a queried landmark"
+  },
+  "rules_present": [
+   "blanket:oxford",
+   "ctx:paris+montmartre",
+   "ctx:paris+notre-dame",
+   "unique:arc de triomphe",
+   "unique:bodleian",
+   "unique:cornmarket",
+   "unique:eiffel",
+   "unique:louvre",
+   "unique:magdalen tower",
+   "unique:moulin rouge",
+   "unique:pompidou",
+   "unique:radcliffe camera",
+   "unique:sacre coeur",
+   "unique:sacre-coeur",
+   "unique:sheldonian"
+  ],
+  "stance": "The matcher is category-name based and deliberately inclusive, without image-level duplicate detection. Verified precision is 43% at image level (13 of 87 classes), varying by rule: unique:<landmark> 58%, blanket:oxford 18%, ctx:* 8%."
+ },
+ "source_artifacts": {
+  "per_class_matches": "fusion-data:/fp_phase2/plan/leakage_report.json",
+  "class_id_list": "fusion-data:/fp_phase3a/plan/leak_class_ids.json",
+  "totals": "fusion-data:/fp_phase3a/plan/plan_full.json"
+ },
+ "classes": [
+  {
+   "landmark_id": 3254,
+   "category": "http://commons.wikimedia.org/wiki/Category:Christ_Church_War_Memorial_Garden,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 3946,
+   "category": "http://commons.wikimedia.org/wiki/Category:St_Barnabas'_Church,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 3957,
+   "category": "http://commons.wikimedia.org/wiki/Category:St_Giles',_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 6190,
+   "category": "http://commons.wikimedia.org/wiki/Category:Radcliffe_Camera",
+   "rule": "unique:radcliffe camera"
+  },
+  {
+   "landmark_id": 19172,
+   "category": "http://commons.wikimedia.org/wiki/Category:All_Souls_College,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 22126,
+   "category": "http://commons.wikimedia.org/wiki/Category:St._Audoen's_Church,_Cornmarket",
+   "rule": "unique:cornmarket"
+  },
+  {
+   "landmark_id": 26614,
+   "category": "http://commons.wikimedia.org/wiki/Category:Carfax,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 27998,
+   "category": "http://commons.wikimedia.org/wiki/Category:The_Codrington_Library,_All_Souls_College,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 28444,
+   "category": "http://commons.wikimedia.org/wiki/Category:Pavillon_Sully_(Louvre)",
+   "rule": "unique:louvre"
+  },
+  {
+   "landmark_id": 28604,
+   "category": "http://commons.wikimedia.org/wiki/Category:Arc_de_Triomphe_du_Carrousel",
+   "rule": "unique:arc de triomphe"
+  },
+  {
+   "landmark_id": 28949,
+   "category": "http://commons.wikimedia.org/wiki/Category:Moulin_Rouge",
+   "rule": "unique:moulin rouge"
+  },
+  {
+   "landmark_id": 32727,
+   "category": "http://commons.wikimedia.org/wiki/Category:%C3%89glise_Saint-Leu-Saint-Gilles_(Chennevi%C3%A8res-l%C3%A8s-Louvres)",
+   "rule": "unique:louvre"
+  },
+  {
+   "landmark_id": 33344,
+   "category": "http://commons.wikimedia.org/wiki/Category:Oxford_Town_Hall",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 35256,
+   "category": "http://commons.wikimedia.org/wiki/Category:Chapelle_Notre-Dame-de-Consolation_(Paris)",
+   "rule": "ctx:paris+notre-dame"
+  },
+  {
+   "landmark_id": 37135,
+   "category": "http://commons.wikimedia.org/wiki/Category:Oxford_University_Museum_of_Natural_History",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 38800,
+   "category": "http://commons.wikimedia.org/wiki/Category:Rue_Montmartre_(Paris)",
+   "rule": "ctx:paris+montmartre"
+  },
+  {
+   "landmark_id": 40605,
+   "category": "http://commons.wikimedia.org/wiki/Category:Pavillon_de_Flore_(Louvre)",
+   "rule": "unique:louvre"
+  },
+  {
+   "landmark_id": 40943,
+   "category": "http://commons.wikimedia.org/wiki/Category:%C3%89glise_du_Sacr%C3%A9-Coeur_de_Rodez",
+   "rule": "unique:sacre-coeur"
+  },
+  {
+   "landmark_id": 44752,
+   "category": "http://commons.wikimedia.org/wiki/Category:Basilique_du_Sacr%C3%A9-Coeur_de_Nancy",
+   "rule": "unique:sacre-coeur"
+  },
+  {
+   "landmark_id": 45029,
+   "category": "http://commons.wikimedia.org/wiki/Category:Macdonald_Randolph_Hotel,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 46593,
+   "category": "http://commons.wikimedia.org/wiki/Category:Wheatley_Mill,_Oxfordshire",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 47378,
+   "category": "http://commons.wikimedia.org/wiki/Category:Eiffel_Tower",
+   "rule": "unique:eiffel"
+  },
+  {
+   "landmark_id": 49907,
+   "category": "http://commons.wikimedia.org/wiki/Category:Bodleian_Library",
+   "rule": "unique:bodleian"
+  },
+  {
+   "landmark_id": 51997,
+   "category": "http://commons.wikimedia.org/wiki/Category:St_Peter's_church,_Wallingford,_Oxfordshire",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 55991,
+   "category": "http://commons.wikimedia.org/wiki/Category:Centre_Pompidou-Metz",
+   "rule": "unique:pompidou"
+  },
+  {
+   "landmark_id": 56627,
+   "category": "http://commons.wikimedia.org/wiki/Category:Cath%C3%A9drale_Notre-Dame_de_Paris",
+   "rule": "ctx:paris+notre-dame"
+  },
+  {
+   "landmark_id": 58639,
+   "category": "http://commons.wikimedia.org/wiki/Category:Louvre-Lens",
+   "rule": "unique:louvre"
+  },
+  {
+   "landmark_id": 62420,
+   "category": "http://commons.wikimedia.org/wiki/Category:Sacre_Coeur_Church_(Saint_Petersburg)",
+   "rule": "unique:sacre coeur"
+  },
+  {
+   "landmark_id": 62571,
+   "category": "http://commons.wikimedia.org/wiki/Category:Christ_Church_Cathedral,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 74264,
+   "category": "http://commons.wikimedia.org/wiki/Category:Oxford_Canal",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 77126,
+   "category": "http://commons.wikimedia.org/wiki/Category:Carfax_Tower,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 80516,
+   "category": "http://commons.wikimedia.org/wiki/Category:%C3%89glise_Notre-Dame-de-Lorette_(Paris)",
+   "rule": "ctx:paris+notre-dame"
+  },
+  {
+   "landmark_id": 83152,
+   "category": "http://commons.wikimedia.org/wiki/Category:High_Street,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 84100,
+   "category": "http://commons.wikimedia.org/wiki/Category:Radcliffe_Square,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 88878,
+   "category": "http://commons.wikimedia.org/wiki/Category:East_facade_of_the_Sully_Wing_(Louvre)",
+   "rule": "unique:louvre"
+  },
+  {
+   "landmark_id": 92774,
+   "category": "http://commons.wikimedia.org/wiki/Category:St_Sepulchre's_Cemetery,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 93486,
+   "category": "http://commons.wikimedia.org/wiki/Category:St_Nicholas'_Church,_Islip,_Oxfordshire",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 94050,
+   "category": "http://commons.wikimedia.org/wiki/Category:Oxford_Circus",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 97721,
+   "category": "http://commons.wikimedia.org/wiki/Category:%C3%89glise_Saint_Justin,_Louvres",
+   "rule": "unique:louvre"
+  },
+  {
+   "landmark_id": 101523,
+   "category": "http://commons.wikimedia.org/wiki/Category:Divinity_School,_Oxford,_England",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 102786,
+   "category": "http://commons.wikimedia.org/wiki/Category:Eiffel_Tower_(Paris,_Texas)",
+   "rule": "unique:eiffel"
+  },
+  {
+   "landmark_id": 103293,
+   "category": "http://commons.wikimedia.org/wiki/Category:Edif%C3%ADcio_Eiffel",
+   "rule": "unique:eiffel"
+  },
+  {
+   "landmark_id": 103894,
+   "category": "http://commons.wikimedia.org/wiki/Category:Selfridges,_Oxford_Street_store",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 104951,
+   "category": "http://commons.wikimedia.org/wiki/Category:Near_Eastern_antiquities_in_the_Louvre",
+   "rule": "unique:louvre"
+  },
+  {
+   "landmark_id": 109280,
+   "category": "http://commons.wikimedia.org/wiki/Category:South_Park,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 112106,
+   "category": "http://commons.wikimedia.org/wiki/Category:St_Giles'_Church,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 114468,
+   "category": "http://commons.wikimedia.org/wiki/Category:%C3%89glise_Notre-Dame-de-Nazareth_(Paris)",
+   "rule": "ctx:paris+notre-dame"
+  },
+  {
+   "landmark_id": 115648,
+   "category": "http://commons.wikimedia.org/wiki/Category:%C3%89glise_Notre-Dame-du-Bon-Conseil_(Paris)",
+   "rule": "ctx:paris+notre-dame"
+  },
+  {
+   "landmark_id": 119086,
+   "category": "http://commons.wikimedia.org/wiki/Category:St_Frideswide's_Church,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 120027,
+   "category": "http://commons.wikimedia.org/wiki/Category:Martyrs'_Memorial,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 121651,
+   "category": "http://commons.wikimedia.org/wiki/Category:Somerville_College,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 124095,
+   "category": "http://commons.wikimedia.org/wiki/Category:%C3%89glise_Notre-Dame-des-Otages_(Paris)",
+   "rule": "ctx:paris+notre-dame"
+  },
+  {
+   "landmark_id": 125112,
+   "category": "http://commons.wikimedia.org/wiki/Category:%C3%89glise_Notre-Dame-du-Rosaire_(Paris)",
+   "rule": "ctx:paris+notre-dame"
+  },
+  {
+   "landmark_id": 127420,
+   "category": "http://commons.wikimedia.org/wiki/Category:Newbridge,_Oxfordshire",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 128748,
+   "category": "http://commons.wikimedia.org/wiki/Category:The_Queen's_College,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 130828,
+   "category": "http://commons.wikimedia.org/wiki/Category:Eiffelgebouw",
+   "rule": "unique:eiffel"
+  },
+  {
+   "landmark_id": 132916,
+   "category": "http://commons.wikimedia.org/wiki/Category:Sheldonian_Theatre",
+   "rule": "unique:sheldonian"
+  },
+  {
+   "landmark_id": 135277,
+   "category": "http://commons.wikimedia.org/wiki/Category:Rue_du_Faubourg-Montmartre_(Paris)",
+   "rule": "ctx:paris+montmartre"
+  },
+  {
+   "landmark_id": 138332,
+   "category": "http://commons.wikimedia.org/wiki/Category:Parvis_Notre-Dame_-_place_Jean-Paul-II_(Paris)",
+   "rule": "ctx:paris+notre-dame"
+  },
+  {
+   "landmark_id": 145137,
+   "category": "http://commons.wikimedia.org/wiki/Category:Basilique_du_Sacr%C3%A9-Coeur_de_Paray-le-Monial",
+   "rule": "unique:sacre-coeur"
+  },
+  {
+   "landmark_id": 145268,
+   "category": "http://commons.wikimedia.org/wiki/Category:Louvre_Pyramid",
+   "rule": "unique:louvre"
+  },
+  {
+   "landmark_id": 147275,
+   "category": "http://commons.wikimedia.org/wiki/Category:Magdalen_Tower",
+   "rule": "unique:magdalen tower"
+  },
+  {
+   "landmark_id": 152496,
+   "category": "http://commons.wikimedia.org/wiki/Category:Christ_Church,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 153186,
+   "category": "http://commons.wikimedia.org/wiki/Category:%C3%89glise_Notre-Dame-du-Travail_(Paris)",
+   "rule": "ctx:paris+notre-dame"
+  },
+  {
+   "landmark_id": 153613,
+   "category": "http://commons.wikimedia.org/wiki/Category:Oxford_College,_Aletheia_University",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 154288,
+   "category": "http://commons.wikimedia.org/wiki/Category:Lady_Margaret_Hall,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 155981,
+   "category": "http://commons.wikimedia.org/wiki/Category:Oxford_Street",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 161310,
+   "category": "http://commons.wikimedia.org/wiki/Category:Ashdown_House,_Oxfordshire",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 162298,
+   "category": "http://commons.wikimedia.org/wiki/Category:%C3%89glise_Notre-Dame-de-l'Assomption_de_Paris",
+   "rule": "ctx:paris+notre-dame"
+  },
+  {
+   "landmark_id": 163046,
+   "category": "http://commons.wikimedia.org/wiki/Category:Ponte_Eiffel",
+   "rule": "unique:eiffel"
+  },
+  {
+   "landmark_id": 163420,
+   "category": "http://commons.wikimedia.org/wiki/Category:Oxford_Terrace_Baptist_Church",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 167275,
+   "category": "http://commons.wikimedia.org/wiki/Category:Bridge_of_Sighs_(Oxford)",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 169267,
+   "category": "http://commons.wikimedia.org/wiki/Category:%C3%89glise_Notre-Dame-de-Bonne-Nouvelle_(Paris)",
+   "rule": "ctx:paris+notre-dame"
+  },
+  {
+   "landmark_id": 170279,
+   "category": "http://commons.wikimedia.org/wiki/Category:Taylor_Institution,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 170812,
+   "category": "http://commons.wikimedia.org/wiki/Category:Temple_protestant_de_l'Oratoire_du_Louvre",
+   "rule": "unique:louvre"
+  },
+  {
+   "landmark_id": 175996,
+   "category": "http://commons.wikimedia.org/wiki/Category:St-Peter-in-the-East,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 181188,
+   "category": "http://commons.wikimedia.org/wiki/Category:New_Theatre_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 183351,
+   "category": "http://commons.wikimedia.org/wiki/Category:%C3%89glise_Notre-Dame_de_Fatima_(Paris)",
+   "rule": "ctx:paris+notre-dame"
+  },
+  {
+   "landmark_id": 183538,
+   "category": "http://commons.wikimedia.org/wiki/Category:Aile_Lescot_(Louvre)",
+   "rule": "unique:louvre"
+  },
+  {
+   "landmark_id": 187280,
+   "category": "http://commons.wikimedia.org/wiki/Category:Buckland_House,_Oxfordshire",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 190111,
+   "category": "http://commons.wikimedia.org/wiki/Category:%C3%89glise_Notre-Dame-de-la-Gare_(Paris)",
+   "rule": "ctx:paris+notre-dame"
+  },
+  {
+   "landmark_id": 190618,
+   "category": "http://commons.wikimedia.org/wiki/Category:Mary_Magdalene,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 193664,
+   "category": "http://commons.wikimedia.org/wiki/Category:St_Bartholomew's_Chapel,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 194122,
+   "category": "http://commons.wikimedia.org/wiki/Category:Wellington_Square,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 194438,
+   "category": "http://commons.wikimedia.org/wiki/Category:Jesus_College,_Oxford",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 195994,
+   "category": "http://commons.wikimedia.org/wiki/Category:Oxford_Castle",
+   "rule": "blanket:oxford"
+  },
+  {
+   "landmark_id": 196183,
+   "category": "http://commons.wikimedia.org/wiki/Category:%C3%89glise_du_Sacr%C3%A9-Coeur_de_Gentilly",
+   "rule": "unique:sacre-coeur"
+  }
+ ]
+}