Dataset format
All the annotations are encoded in JSON files. Two examples showing the annotations for the images below are presented. A detailed description of each entry from the JSON file is provided in the original paper.
{
"form": [
{
"id": 0,
"rect": true,
"box": [
279,
36,
412,
83
],
"text": "MAGAZINE SCORES AUDIENCE STUDIES",
"lines": [
{
"box": [
279,
36,
404,
53
],
"text": "MAGAZINE SCORES",
"words": [
{
"box": [
279,
36,
345,
53
],
"text": "MAGAZINE"
},
{
"box": [
351,
36,
404,
53
],
"text": "SCORES"
}
]
},
{
"box": [
281,
63,
412,
83
],
"text": "AUDIENCE STUDIES",
"words": [
{
"box": [
281,
63,
347,
83
],
"text": "AUDIENCE"
},
{
"box": [
352,
64,
412,
81
],
"text": "STUDIES"
}
]
}
],
"label": "header",
"linking": [
[
0,
1
],
[
0,
3
]
]
},
{
"id": 1,
"rect": true,
"box": [
41,
127,
84,
144
],
"text": "Brand",
"lines": [
{
"box": [
41,
127,
84,
144
],
"text": "Brand",
"words": [
{
"box": [
41,
127,
84,
144
],
"text": "Brand"
}
]
}
],
"label": "question",
"linking": [
[
0,
1
],
[
1,
2
]
]
},
{
"id": 2,
"rect": true,
"box": [
138,
124,
236,
143
],
"text": "SALEM (RJR)",
"lines": [
{
"box": [
138,
124,
236,
143
],
"text": "SALEM (RJR)",
"words": [
{
"text": "SALEM",
"box": [
138,
126,
184,
143
]
},
{
"text": "(RJR)",
"box": [
194,
124,
236,
142
]
}
]
}
],
"label": "answer",
"linking": [
[
1,
2
]
]
},
{
"id": 3,
"rect": true,
"box": [
459,
127,
538,
145
],
"text": "Project #",
"lines": [
{
"box": [
459,
127,
538,
145
],
"text": "Project #",
"words": [
{
"text": "Project",
"box": [
459,
127,
519,
144
]
},
{
"text": "#",
"box": [
525,
127,
538,
145
]
}
]
}
],
"label": "question",
"linking": [
[
0,
3
],
[
3,
4
]
]
},
{
"id": 4,
"rect": true,
"box": [
568,
123,
612,
141
],
"text": "74- 80",
"lines": [
{
"box": [
568,
123,
612,
141
],
"text": "74- 80",
"words": [
{
"text": "74-",
"box": [
568,
123,
595,
141
]
},
{
"text": "80",
"box": [
594,
123,
612,
138
]
}
]
}
],
"label": "answer",
"linking": [
[
3,
4
]
]
}
]
}
The following is an example of annotation for an item table, where the entity IDs correspond to the complete annotation JSON.
{
"form": [
{
"table_name": "item_table_0",
"entity_ids": [
22,
23,
24,
25,
26,
27,
28,
29,
30,
31,
32,
33,
34,
35,
36,
37,
38,
39,
40,
41,
42,
43,
44,
45
],
"table_box": [
45,
318,
654,
615
],
"item_value_ids": {
"item_table_0_0": [
26,
27,
28,
29
],
"item_table_0_1": [
32,
33,
30,
31
],
"item_table_0_2": [
34,
35,
36,
37
],
"item_table_0_3": [
40,
41,
38,
39
],
"item_table_0_4": [
42,
43,
44,
45
]
},
"item_value_boxes": {
"item_table_0_0": [
45,
338,
612,
372
],
"item_table_0_1": [
45,
391,
610,
458
],
"item_table_0_2": [
45,
476,
612,
511
],
"item_table_0_3": [
47,
531,
610,
561
],
"item_table_0_4": [
46,
582,
611,
615
]
}
}
]
}