Ibnuck/poster-schedule-information-extraction
Multimodal Visual-Text Dataset for Poster Schedule Information Extraction A ready-to-train Indonesian document AI dataset combining pixels, OCR tokens, spatial layout, and BIO entity labels. Overview What it is 127 Indonesian seminar and religious-study event posters with multimodal token-level annotations Primary task Schedule information extraction as token classification Modalities Image + text + 2D spatial layout Coordinates… See the full description on the dataset page: https://huggingface.co/datasets/Ibnuck/poster-schedule-information-extraction.
1115
1{2 "seed": 42,3 "strategy": "rare-label-aware split reproduced from Training.ipynb",4 "rare_label_threshold": 10,5 "rare_labels": [6 "B-TANGGALLENGKAP"7 ],8 "splits": {9 "train": [10 "9",11 "91",12 "101",13 "77",14 "49",15 "16",16 "63",17 "102",18 "103",19 "82",20 "47",21 "21",22 "30",23 "86",24 "32",25 "50",26 "64",27 "123",28 "116",29 "18",30 "75",31 "2",32 "67",33 "80",34 "36",35 "85",36 "7",37 "1",38 "117",39 "79",40 "26",41 "98",42 "61",43 "74",44 "94",45 "60",46 "39",47 "73",48 "34",49 "51",50 "62",51 "38",52 "110",53 "41",54 "99",55 "104",56 "56",57 "42",58 "83",59 "22",60 "52",61 "71",62 "23",63 "96",64 "40",65 "105",66 "120",67 "68",68 "6",69 "55",70 "14",71 "89",72 "66",73 "97",74 "106",75 "93",76 "8",77 "24",78 "46",79 "37",80 "10",81 "81",82 "15",83 "69",84 "58",85 "5",86 "33",87 "44",88 "45",89 "12",90 "48",91 "108",92 "118",93 "88",94 "107",95 "19",96 "100",97 "43",98 "111",99 "20",100 "0",101 "124",102 "112",103 "57",104 "122",105 "53",106 "114",107 "90",108 "84",109 "92",110 "121"111 ],112 "validation": [113 "25",114 "72",115 "109",116 "78",117 "65",118 "29",119 "27",120 "113",121 "126",122 "4",123 "54",124 "76"125 ],126 "test": [127 "11",128 "70",129 "115",130 "119",131 "87",132 "13",133 "125",134 "17",135 "28",136 "31",137 "35",138 "95",139 "3",140 "59"141 ]142 }143}