GQA数据集结构的详细描述

pinkshell_1314

已于 2023-10-10 19:57:34 修改

阅读量571

点赞数 2

文章标签：计算机视觉人工智能

于 2023-03-22 14:11:27 首次发布

本文链接：https://blog.csdn.net/pinkshell_1314/article/details/129637513

版权

官网地址：GQA: Visual Reasoning in the Real World (stanford.edu)

一、数据集结构

GQA数据集由三个部分组成：场景图、问题和图像。场景图包含了训练集和验证集中每一张图像的场景图信息；问题包含了回答这个问题所需要的一些推理步骤；图像这个文件夹下面除了有图片，还储存了每张图像中包含的所有对象的特征和每张图像的空间特征。

二、数据集详解

1.allImages

（1）概述

（2）images

存储原始图像的文件夹。

图片数：148854

（3）spatial

存储图片的空间特征。

1> gqa_spatial_0.h5

10000个7*7*2048维的特征。

2> gqa_spatial_info.json

存储相对应的特征，图像编号，以及所在文件位置和索引的字典。

{key：“图片编号” value： {key：“idx” value：h5文件中图像的索引， key：“file” value:所在文件的索引} }

图片编号：1-2417997(比较连续)，n1-n581923（不连续）

特征编号（idx）：一般 0-9999，10号文件0-8078，15号文件 0-774

（4）objects

存储图片里的对象特征。

1> gqa_objects_0.h5

分为两个部分：bboxes,features

① bboxes:10000个4*1*100维向量，每一个向量保存一张图像上的所有对象的包围框顶点的坐标

记录每一个框的 x,y,w,h

② features

10000个2048*1*100维向量

保存每个目标的特征信息

2> gqa_objects_info.json

存储相对应的图像编号，图像中的对象数，图像的宽度和高度以及所在文件位置和索引的字典。

{key：“图片编号” value： {key：“width” value：图像的宽度，key：“objectsNum” value：图片中的对象数，key：“idx” value：h5文件中图像的索引，key：“height” value：图像的高度， key：“file” value:所在文件的索引} }

特征编号（idx）：一般 0-9999，10号文件0-8078，15号文件 0-774

2.sceneGraphs

1>train_sceneGraphs.json

{key：图像编号 value：{key：“weather” value：图像里显示的场景的天气，key：“width” value：图像宽度，key：“objects” value：{key：对象编号 value：{key：“name” value：对象名称，key：“h” value：对象高度，key：“relations” value：[{key：“objects” value：对象编号，key：“name” value：对象之间的关系}{key：“objects” value：对象编号，key：“name” value：对象之间的关系}]key：“w” value：对象的宽度，key：“attributes” value:[对象的属性列表]，key：“y” value：对象左上角的y坐标，key：“x” value：对象左上角的x坐标}}key：“location” value：图像里显示的其所属的位置，key：“height” value：图像的高度}}

3. questions1.2

1> train_all_questions/train_all_questions_0.json

{key：问题ID value：{key：“semantic” value：[key：“operation” value：推理操作，key：“dependencies” value：当前步骤所依赖的先前步骤，key：“argument” value：操作参数（取决于具体操作，通常是对象ID）] key：“entailed” value：包含的问题的问题ID列表，key：“equivalent” value：等效问题的问题ID列表，key：“question” value：问题，key：“imgageId” value：问题对应的图片ID，key：“isBalanced” value：是否经过平衡，key：“groups” value：{key：“global” value：问题全局组代码(例如，所有颜色的问题)，key：“local” value：问题本地组代码（例如，有关苹果颜色的所有问题)}key：“answer” value：简短答案，key：“semanticStr” value：问题语义结构的字符串形式，key：“annotations” value：{key：“answer” value：从答案词（例如索引“0”，key）到对象（对象 ID，value）的视觉指针，key：“question” value：从疑问词（例如切片“2：4”，key）到对象（对象 Id，value）的视觉指针，key：“fullAnswer” value：{从回答词（例如：“0”、“2：4”、key）到对象（objectId、value）的视觉指针}key：“types” value：{key：“detailed” value：该问题完整类型规范，key：“semantic” value：问题主题的类型，key：“structural” value：问题结构类型}key：“fullAnswer” value：长版答案}}

2>testdev_all_questions.json

（测试集）

3>test_all_questions.json

（只保留了图像和问题的关联信息以及平衡信息用于测试）

三、一个样本的完整结构

1.图像

路径： images/2266.jpg

2.空间特征

路径：spatial/gqa_spatial_info.json

"2266": {"idx": 2265, "file": 0}

路径：spatial/gqa_spatial_0.h5

3. 对象特征

路径：objects/gqa_objects_info.json

"2266": {    # 图片编号
    "width": 800,     # 图像的宽度
    "objectsNum": 69,    # 图片中的对象数
    "idx": 3901,     # h5文件中图像的索引
    "height": 600,     # 图像的高度
    "file": 0    # 所在文件的索引
    }

路径：objects/gqa_objects_0.h5

4.场景图

路径：sceneGraphs/train_sceneGraphs.json

描述：x,y是对象框左上顶点，w,h是对象框的宽和高

"2266": { # 图像编号
"width": 800, # 图像宽度
"objects": { # 从对象ID到其对象的字典

        "4324200": { # 对象编号ID
            "name": "person", # 对象名称
             "h": 70, # 对象高度
             "relations": [ # 对象（源）的所有传出关系（边）的列表
                {
           "object": "4324184", # 对象编号
           "name": "to the left of" # 对象之间的关系
           },
                {"object": "4324192", "name": "to the left of"},
                {"object": "4324213", "name": "to the left of"},
                {"object": "4324196", "name": "to the left of"},
                {"object": "4324219", "name": "to the left of"},
                {"object": "4324190", "name": "to the right of"},
                {"object": "4537542", "name": "to the left of"},
                {"object": "4324189", "name": "to the right of"},
                {"object": "4324188", "name": "to the left of"},
                {"object": "4324191", "name": "to the right of"}
                    ],
            "w": 20,   # 对象的宽度
            "attributes": [], # 对象的属性列表
            "y": 388,   # 对象左上角的y坐标
            "x": 85 # 对象左上角的x坐标
                },

       "4537543": {
           "name": "crosswalk",
           "h": 158,
           "relations": [
               {"object": "4537536", "name": "to the left of"}
                   ],
           "w": 453,
           "attributes": [],
           "y": 436,
           "x": 12
               },

       "4537542": {
           "name": "people",
           "h": 102,
           "relations": [
               {"object": "4324213", "name": "to the right of"},
               {"object": "4324191", "name": "to the right of"},
               {"object": "4324184", "name": "to the left of"},
               {"object": "4324188", "name": "to the left of"},
               {"object": "4324190", "name": "to the right of"},
               {"object": "4324192", "name": "to the left of"},
               {"object": "4324200", "name": "to the right of"},
               {"object": "4324189", "name": "to the right of"},
               {"object": "4324187", "name": "to the left of"}],
           "w": 34,
           "attributes": [],
           "y": 384,
           "x": 141
               },

        "4324187": {
           "name": "bags",
           "h": 38,
           "relations": [
               {"object": "4324213", "name": "to the right of"},
               {"object": "4537542", "name": "to the right of"},
               {"object": "4324189", "name": "to the right of"},
               {"object": "4324184", "name": "to the left of"},
               {"object": "4324190", "name": "to the right of"}
                   ],
           "w": 14,
           "attributes": [],
           "y": 447,
           "x": 180
               },

       "4324183": {
           "name": "road",
           "h": 167,
           "relations": [],
           "w": 795,
           "attributes": [],
           "y": 427,
           "x": 0
               },

       "4324189": {
           "name": "man",
           "h": 156,
           "relations": [
               {"object": "4324213", "name": "to the left of"},
               {"object": "4324196", "name": "to the left of"},
               {"object": "4324192", "name": "to the left of"},
               {"object": "4324219", "name": "to the left of"},
               {"object": "4324188", "name": "to the left of"},
               {"object": "4324187", "name": "to the left of"},
               {"object": "4537542", "name": "to the left of"},
               {"object": "4324200", "name": "to the left of"},
               {"object": "4324184", "name": "to the left of"},
               {"object": "4324190", "name": "with"}
                   ],
           "w": 58,
           "attributes": [],
           "y": 354,
           "x": 31
               },

        "4324188": {
           "name": "hand",
           "h": 10,
           "relations": [
               {"object": "4324190", "name": "to the right of"},
               {"object": "4324213", "name": "to the right of"},
               {"object": "4324189", "name": "to the right of"},
               {"object": "4324192", "name": "to the left of"},
               {"object": "4324200", "name": "to the right of"},
               {"object": "4537542", "name": "to the right of"}
                   ],
           "w": 12,
           "attributes": [],
           "y": 434,
           "x": 188
               },

        "4324219": {
           "name": "window",
           "h": 131,
           "relations": [
               {"object": "4324211", "name": "to the right of"},
               {"object": "4324191", "name": "to the right of"},
               {"object": "4537539", "name": "to the left of"},
               {"object": "4324200", "name": "to the right of"},
                {"object": "4324189", "name": "to the right of"},
               {"object": "4324212", "name": "to the right of"}
                   ],
           "w": 183,
           "attributes": ["arched"],
           "y": 296,
           "x": 270
               },

       "4324184": {
           "name": "people",
           "h": 158,
           "relations": [
               {"object": "4537542", "name": "to the right of"},
               {"object": "4324213", "name": "to the right of"},
               {"object": "4537536", "name": "to the left of"},
               {"object": "4324187", "name": "to the right of"},
               {"object": "4324189", "name": "to the right of"},
               {"object": "4324190", "name": "to the right of"},
               {"object": "4324191", "name": "to the right of"},
               {"object": "4324200", "name": "to the right of"},
               {"object": "4324193", "name": "to the left of"},
               {"object": "4324192", "name": "to the left of"}
                   ],
           "w": 66,
           "attributes": [],
           "y": 357,
           "x": 192
               },

       "4324192": {
           "name": "car",
           "h": 74,
           "relations": [
               {"object": "4324189", "name": "to the right of"},
               {"object": "4537542", "name": "to the right of"},
               {"object": "4324200", "name": "to the right of"},
               {"object": "4324188", "name": "to the right of"},
               {"object": "4537536", "name": "to the left of"},
               {"object": "4324184", "name": "to the right of"},
                {"object": "4324190", "name": "to the right of"},
               {"object": "4324191", "name": "to the right of"},
               {"object": "4324213", "name": "to the right of"}
                   ],
            "w": 228,
           "attributes": [],
           "y": 392,
           "x": 279
               },

       "4324193": {
           "name": "man",
           "h": 71,
           "relations": [
               {"object": "4537539", "name": "to the left of"},
               {"object": "4324184", "name": "to the right of"},
               {"object": "4537536", "name": "to the left of"}],
           "w": 26,
           "attributes": [],
           "y": 386,
           "x": 400
               },

       "4324190": {
           "name": "jeans",
           "h": 85,
           "relations": [
               {"object": "4324213", "name": "to the left of"},
               {"object": "4324192", "name": "to the left of"},
               {"object": "4324196", "name": "to the left of"},
               {"object": "4324187", "name": "to the left of"},
               {"object": "4537542", "name": "to the left of"},
               {"object": "4324200", "name": "to the left of"},
               {"object": "4324184", "name": "to the left of"},
               {"object": "4324188", "name": "to the left of"}
                   ],
           "w": 54,
           "attributes": [],
           "y": 419,
           "x": 35
               },

        "4324191": {
           "name": "shirt",
           "h": 45,
           "relations": [
               {"object": "4537542", "name": "to the left of"},
               {"object": "4324184", "name": "to the left of"},
               {"object": "4324200", "name": "to the left of"},
               {"object": "4324219", "name": "to the left of"},
               {"object": "4324192", "name": "to the left of"},
               {"object": "4324213", "name": "to the left of"}
                   ],
           "w": 56,
           "attributes": ["white"],
           "y": 375,
           "x": 33
               },

       "4324196": {
           "name": "sidewalk",
           "h": 43,
           "relations": [
               {"object": "4324189", "name": "to the right of"},
               {"object": "4324200", "name": "to the right of"},
               {"object": "4324190", "name": "to the right of"}
                   ],
           "w": 487,
           "attributes": [],
           "y": 409,
           "x": 107
               },

       "4324211": {
           "name": "traffic sign",
           "h": 16,
           "relations": [
               {"object": "4324219", "name": "to the left of"},
               {"object": "4324212", "name": "on"}
                   ],
           "w": 42,
           "attributes": [],
           "y": 298,
           "x": 119
               },

        "4324210": {
           "name": "building",
           "h": 448,
           "relations": [],
           "w": 779,
           "attributes": [],
           "y": 0,
           "x": 11
               },

      "4324213": {
           "name": "people",
           "h": 114,
           "relations": [
               {"object": "4324191", "name": "to the right of"},
               {"object": "4324188", "name": "to the left of"},
               {"object": "4324184", "name": "to the left of"},
               {"object": "4324192", "name": "to the left of"},
               {"object": "4324190", "name": "to the right of"},
               {"object": "4324200", "name": "to the right of"},
               {"object": "4537542", "name": "to the left of"},
               {"object": "4324187", "name": "to the left of"},
               {"object": "4324189", "name": "to the right of"},
                {"object": "4324214", "name": "walking in"}
                   ],
           "w": 42,
           "attributes": [],
           "y": 377,
           "x": 98
               },

       "4324212": {
           "name": "pole",
           "h": 80,
           "relations": [
               {"object": "4324219", "name": "to the left of"}
                   ],
           "w": 17,
           "attributes": [],
           "y": 313,
           "x": 130
               },

       "4324214": {
           "name": "intersection",
           "h": 269,
           "relations": [
               {"object": "4537539", "name": "to the left of"},
               {"object": "4537536", "name": "to the left of"}
                   ],
           "w": 517,
           "attributes": [],
           "y": 328,
           "x": 5
               },

       "4537538": {
           "name": "street",
           "h": 145,
           "relations": [],
           "w": 735,
           "attributes": [],
           "y": 449,
           "x": 39
               },

       "4537539": {
           "name": "van",
           "h": 59,
           "relations": [
               {"object": "4537538", "name": "on"},
               {"object": "4324214", "name": "to the right of"},
               {"object": "4324196", "name": "parked near"},
               {"object": "4324193", "name": "to the right of"},
               {"object": "4324219", "name": "to the right of"}
                   ],
            "w": 121,
            "attributes": ["white"],
           "y": 364,
           "x": 623
               },

       "4537536": {
           "name": "car",
           "h": 136,
           "relations": [
               {"object": "4324184", "name": "to the right of"},
               {"object": "4537543", "name": "to the right of"},
               {"object": "4324193", "name": "to the right of"},
               {"object": "4324192", "name": "to the right of"},
               {"object": "4324214", "name": "to the right of"}
                   ],
            "w": 336,
            "attributes": ["black"],
           "y": 384,
           "x": 414
               }

    "location": "outdoors",   # 图像里显示的其所属的位置（室内室外）
    "height": 600 # 图像的高度
    },

5. 问题和答案

（1）全部版本（224）

路径：questions1.2/train_all_questions/train_all_questions_0.json（39个）

"1123309": {    # 问题ID
    "semantic": [    # 回答问题所需的推理步骤列表
        {
            "operation": "select",    # 推理操作
            "dependencies": [],    # 当前步骤所依赖的先前步骤
            "argument": "bag (4324187)"    # 操作参数。取决于具体操作，通常是对象ID
         },
        {
            "operation": "filter hposition",
            "dependencies": [0],
            "argument": "left"
         },
        {
            "operation": "relate",
            "dependencies": [1],
            "argument": "apple,to the left of,s (-)"
        },
        {
            "operation": "exist",
            "dependencies": [2],
            "argument": "?"
        }
    ],
    "entailed": ["1123308", "1123310"],    # 包含的问题的问题ID列表
    "equivalent": ["1123309", "1123310"],    # 等效问题的问题ID列表
    "question": "Are there apples to the left of the bags on the left?",    # 问题
    "imageId": "2266",    # 问题对应的图片ID
    "isBalanced": false,    # 是否经过平衡
    "groups": {    # 问题组代码，用于平衡数据集
        "global": null,    # 问题全局组代码(例如，所有颜色的问题)
        "local": "13-bags_apple"    # 问题本地组代码（例如，有关苹果颜色的所有问题)
             },
    "answer": "no",    # 简短答案
    "semanticStr": "select: bag (4324187)->filter hposition: left [0]->relate: apple,to the left of,s (-) [1]->exist: ? [2]",    # 问题语义结构的字符串形式
    "annotations": {    # 问题和答案的对象批注
        "answer": {},    # 从答案词（例如索引“0”，key）到对象（对象 ID，value）的视觉指针
        "question": {"8": "4324187"},    # 从疑问词（例如切片“2：4”，key）到对象（对象 Id，value）的视觉指针
        "fullAnswer": {    # 从回答词（例如：“0”、“2：4”、key）到对象（objectId、value）的视觉指针
            "10": "4324187",
            "4": "4324189"
                       }
                    },
    "types": {    # 保存有关问题类型的信息的字典
        "detailed": "existRelSC",    # 该问题完整类型规范
        "semantic": "rel",    # 问题主题的类型
        "structural": "verify"    # 问题结构类型
            },
    "fullAnswer": "No, there is a man to the left of the bags."    # 长版答案
},

"1123308": {
    "semantic": [
        {"operation": "select", "dependencies": [], "argument": "bags (4324187)"},
        {"operation": "relate", "dependencies": [0], "argument": "apple,to the left of,s (-)"},
        {"operation": "exist", "dependencies": [1], "argument": "?"}
    ],
    "entailed": ["1123309", "1123310"],
    "equivalent": ["1123308"],
    "question": "Do you see an apple to the left of the bags?",
    "imageId": "2266",
    "isBalanced": false,
    "groups": {"global": null, "local": "13-bags_apple"},
    "answer": "no",
    "semanticStr": "select: bags (4324187)->relate: apple,to the left of,s (-) [0]->exist: ? [1]",
    "annotations": {"answer": {}, "question": {"10": "4324187"},"fullAnswer": {"10": "4324187", "4": "4324189"}},
    "types": {"detailed": "existRelSC", "semantic": "rel", "structural": "verify"},
    "fullAnswer": "No, there is a man to the left of the bags."
},
"1123301": {
    "semantic": [
        {"operation": "select", "dependencies": [], "argument": "car (4324192)"},
        {"operation": "filter vposition", "dependencies": [0], "argument": "bottom"},
        {"operation": "relate", "dependencies": [1], "argument": "truck,to the left of,s (-)"},
        {"operation": "exist", "dependencies": [2], "argument": "?"}
    ],
    "entailed": ["1123300", "1123302"],
    "equivalent": ["1123301", "1123300", "1123302"],
    "question": "Are there trucks to the left of the car in the bottom?",
    "imageId": "2266",
    "isBalanced": false,
    "groups": {"global": null, "local": "13-car_truck"},
    "answer": "no",
    "semanticStr": "select: car (4324192)->filter vposition: bottom [0]->relate: truck,to the left of,s (-) [1]->exist: ? [2]",
    "annotations": {"answer": {},"question": {"8": "4324192"}, "fullAnswer": {"10": "4324192", "4": "4324189"}},
    "types": {"detailed": "existRelSC", "semantic": "rel", "structural": "verify"},
    "fullAnswer": "No, there is a man to the left of the car."
},
"1123300": {
    "semantic": [
        {"operation": "select", "dependencies": [], "argument": "car (4324192)"},
        {"operation": "filter vposition", "dependencies": [0], "argument": "bottom"},
        {"operation": "relate", "dependencies": [1], "argument": "truck,to the left of,s (-)"},
        {"operation": "exist", "dependencies": [2], "argument": "?"}
    ],
    "entailed": ["1123301", "1123302"],
    "equivalent": ["1123301", "1123300", "1123302"],
    "question": "Is there a truck to the left of the car in the bottom part of the photo?",
    "imageId": "2266",
    "isBalanced": false,
    "groups": {"global": null, "local": "13-car_truck"},
    "answer": "no",
    "semanticStr": "select: car (4324192)->filter vposition: bottom [0]->relate: truck,to the left of,s (-) [1]->exist: ? [2]",
    "annotations": {"answer": {}, "question": {"9": "4324192"}, "fullAnswer": {"10": "4324192", "4": "4324189"}},
    "types": {"detailed": "existRelSC", "semantic": "rel", "structural": "verify"},
    "fullAnswer": "No, there is a man to the left of the car."
},
"1123303": {
    "semantic": [
        {"operation": "select", "dependencies": [], "argument": "bags (4324187)"},
        {"operation": "relate", "dependencies": [0], "argument": "man,to the left of,s (4324189)"},
        {"operation": "exist", "dependencies": [1], "argument": "?"}
    ],
    "entailed": ["1123305", "1123304", "1123307", "1123306"],
    "equivalent": ["1123303"],
    "question": "Do you see men to the left of the bags?",
    "imageId": "2266",
    "isBalanced": false,
    "groups": {"global": null, "local": "13-bags_man"},
    "answer": "yes",
    "semanticStr": "select: bags (4324187)->relate: man,to the left of,s (4324189) [0]->exist: ? [1]",
    "annotations": {"answer": {}, "question": {"9": "4324187", "3": "4324189"}, "fullAnswer": {"10": "4324187", "4": "4324189"}},
    "types": {"detailed": "existRelS", "semantic": "rel", "structural": "verify"},
    "fullAnswer": "Yes, there is a man to the left of the bags."
},
"1123302": {
    "semantic": [
        {"operation": "select", "dependencies": [], "argument": "car (4324192)"},
        {"operation": "filter vposition", "dependencies": [0], "argument": "bottom"},
        {"operation": "relate", "dependencies": [1], "argument": "truck,to the left of,s (-)"},
        {"operation": "exist", "dependencies": [2], "argument": "?"}
    ],
    "entailed": ["1123301", "1123300"],
    "equivalent": ["1123301", "1123300", "1123302"],
    "question": "Do you see a truck to the left of the car that is in the bottom?",
    "imageId": "2266",
    "isBalanced": false,
    "groups": {"global": null, "local": "13-car_truck"},
    "answer": "no",
    "semanticStr": "select: car (4324192)->filter vposition: bottom [0]->relate: truck,to the left of,s (-) [1]->exist: ? [2]",
    "annotations": {"answer": {}, "question": {"10": "4324192"}, "fullAnswer": {"10": "4324192", "4": "4324189"}},
    "types": {"detailed": "existRelSC", "semantic": "rel", "structural": "verify"},
    "fullAnswer": "No, there is a man to the left of the car."
},
"1123305": {
    "semantic": [
        {"operation": "select", "dependencies": [], "argument": "people (4324184)"},
        {"operation": "relate", "dependencies": [0], "argument": "bag,to the left of,s (4324187)"},
        {"operation": "relate", "dependencies": [1], "argument": "man,to the left of,s (4324189)"},
        {"operation": "exist", "dependencies": [2], "argument": "?"}
    ],
    "entailed": ["1123303", "1123304", "1123307", "1123306"],
    "equivalent": ["1123305", "1123304"],
    "question": "Do you see men to the left of the bags which are to the left of the people?",
    "imageId": "2266",
    "isBalanced": false,
    "groups": {"global": null, "local": "13-bags_man"},
    "answer": "yes",
    "semanticStr": "select: people (4324184)->relate: bag,to the left of,s (4324187) [0]->relate: man,to the left of,s (4324189) [1]->exist: ? [2]",
    "annotations": {"answer": {}, "question": {"9": "4324187", "3": "4324189", "17": "4324184"}, "fullAnswer": {"10": "4324187", "4": "4324189"}},
    "types": {"detailed": "existRelS", "semantic": "rel", "structural": "verify"},
    "fullAnswer": "Yes, there is a man to the left of the bags."
},
"1123304": {
    "semantic": [
        {"operation": "select", "dependencies": [], "argument": "people (4324184)"},
        {"operation": "relate", "dependencies": [0], "argument": "bag,to the left of,s (4324187)"},
        {"operation": "relate", "dependencies": [1], "argument": "man,to the left of,s (4324189)"},
        {"operation": "exist", "dependencies": [2], "argument": "?"}
    ],
    "entailed": ["1123303", "1123305", "1123307", "1123306"],
    "equivalent": ["1123305", "1123304"],
    "question": "Are there any men to the left of the bags which are to the left of the people?",
    "imageId": "2266",
    "isBalanced": false,
    "groups": {"global": null, "local": "13-bags_man"},
    "answer": "yes",
    "semanticStr": "select: people (4324184)->relate: bag,to the left of,s (4324187) [0]->relate: man,to the left of,s (4324189) [1]->exist: ? [2]",
    "annotations": {"answer": {}, "question": {"9": "4324187", "3": "4324189", "17": "4324184"}, "fullAnswer": {"10": "4324187", "4": "4324189"}},
    "types": {"detailed": "existRelS", "semantic": "rel", "structural": "verify"},
    "fullAnswer": "Yes, there is a man to the left of the bags."
},
"1123307": {
    "semantic": [
        {"operation": "select", "dependencies": [], "argument": "people (4324184)"},
        {"operation": "relate", "dependencies": [0], "argument": "bag,to the left of,s (4324187)"},
        {"operation": "relate", "dependencies": [1], "argument": "man,to the right of,s (4324189)"},
        {"operation": "exist", "dependencies": [2], "argument": "?"}
    ],
    "entailed": ["1123306"],
    "equivalent": ["1123307"],
    "question": "Are there any men to the right of the bags that are to the left of the people?",
    "imageId": "2266",
    "isBalanced": false,
    "groups": {"global": null, "local": "13-bags_man"},
    "answer": "no",
    "semanticStr": "select: people (4324184)->relate: bag,to the left of,s (4324187) [0]->relate: man,to the right of,s (4324189) [1]->exist: ? [2]",
    "annotations": {"answer": {}, "question": {"9": "4324187", "3": "4324189", "17": "4324184"}, "fullAnswer": {"9": "4324187", "2": "4324189"}},
    "types": {"detailed": "existRelSRC", "semantic": "rel", "structural": "verify"},
    "fullAnswer": "No, the man is to the left of the bags."},
"1123306": {
    "semantic": [
        {"operation": "select", "dependencies": [], "argument": "bags (4324187)"},
        {"operation": "relate", "dependencies": [0], "argument": "man,to the right of,s (4324189)"},
        {"operation": "exist", "dependencies": [1], "argument": "?"}
    ],
    "entailed": ["1123307"],
    "equivalent": ["1123306"],
    "question": "Do you see any men to the right of the bags?",
    "imageId": "2266",
    "isBalanced": false,
    "groups": {"global": null, "local": "13-bags_man"},
    "answer": "no",
    "semanticStr": "select: bags (4324187)->relate: man,to the right of,s (4324189) [0]->exist: ? [1]",
    "annotations": {"answer": {}, "question": {"10": "4324187", "4": "4324189"}, "fullAnswer": {"9": "4324187", "2": "4324189"}},
    "types": {"detailed": "existRelSRC", "semantic": "rel", "structural": "verify"},
    "fullAnswer": "No, the man is to the left of the bags."
},

"1123310": {"semantic": [{"operation": "select", "dependencies": [], "argument": "bag (4324187)"}, {"operation": "filter hposition", "dependencies": [0], "argument": "left"}, {"operation": "relate", "dependencies": [1], "argument": "apple,to the left of,s (-)"}, {"operation": "exist", "dependencies": [2], "argument": "?"}], "entailed": ["1123309", "1123308"], "equivalent": ["1123309", "1123310"], "question": "Do you see any apples to the left of the bags on the left?", "imageId": "2266", "isBalanced": false, "groups": {"global": null, "local": "13-bags_apple"}, "answer": "no", "semanticStr": "select: bag (4324187)->filter hposition: left [0]->relate: apple,to the left of,s (-) [1]->exist: ? [2]", "annotations": {"answer": {}, "question": {"10": "4324187"}, "fullAnswer": {"10": "4324187", "4": "4324189"}}, "types": {"detailed": "existRelSC", "semantic": "rel", "structural": "verify"}, "fullAnswer": "No, there is a man to the left of the bags."},

路径：questions1.2/train_all_questions/train_all_questions_1.json（16）

路径：questions1.2/train_all_questions/train_all_questions_2.json（17）

路径：questions1.2/train_all_questions/train_all_questions_3.json（36）

路径：questions1.2/train_all_questions/train_all_questions_4.json（49）

路径：questions1.2/train_all_questions/train_all_questions_5.json（29）

路径：questions1.2/train_all_questions/train_all_questions_6.json（1）

路径：questions1.2/train_all_questions/train_all_questions_7.json（14）

路径：questions1.2/train_all_questions/train_all_questions_8.json（3）

路径：questions1.2/train_all_questions/train_all_questions_9.json（20）

（2）平衡版本（9）

路径：questions1.2/train_balanced_questions.json

"1123347": {
    "semantic": [
        {"operation": "select","dependencies": [], "argument": "intersection (4324214)"}, 
        { "operation": "relate", "dependencies": [0],"argument": "people,walking in,s (4324213)"}, 
        {"operation": "relate", "dependencies": [1], "argument": "car,to the left of,s (4324192)"}, 
        {"operation": "exist", "dependencies": [2], "argument": "?"}
                ], 
    "entailed": ["1123348"], 
    "equivalent": ["1123347", "1123348"], 
    "question": "Are there any cars to the left of the people that are walking in the intersection?", 
    "imageId": "2266", 
    "isBalanced": true, 
    "groups": {"global": null, "local": "13-people_car"}, 
    "answer": "no", 
    "semanticStr": "select: intersection (4324214)->relate: people,walking in,s (4324213) [0]->relate: car,to the left of,s (4324192) [1]->exist: ? [2]", 
    "annotations": {
        "answer": {}, 
        "question": {"9": "4324213", "3": "4324192", "15": "4324214"},
        "fullAnswer": {"9": "4324213", "2": "4324192"}
    },
    "types": {"detailed": "existRelSRC", "semantic": "rel", "structural": "verify"}, 
    "fullAnswer": "No, the car is to the right of the people."},

"1123263": {"semantic": [{"operation": "select", "dependencies": [], "argument": "man (4324189)"}, {"operation": "relate", "dependencies": [0], "argument": "shirt,with,o (4324191)"}, {"operation": "relate", "dependencies": [1], "argument": "vehicle,same color,_ (4537539)"}, {"operation": "query", "dependencies": [2], "argument": "name"}], "entailed": [], "equivalent": ["1123263"], "question": "What is the vehicle that has the same color as the shirt the man is with called?", "imageId": "2266", "isBalanced": true, "groups": {"global": "vehicle", "local": "14-shirt_same color,_"}, "answer": "van", "semanticStr": "select: man (4324189)->relate: shirt,with,o (4324191) [0]->relate: vehicle,same color,_ (4537539) [1]->query: name [2]", "annotations": {"answer": {"0": "4537539"}, "question": {"11": "4324191", "3": "4537539", "13": "4324189"}, "fullAnswer": {"1": "4537539", "4": "4537539"}}, "types": {"detailed": "sameRelate", "semantic": "rel", "structural": "query"}, "fullAnswer": "The vehicle is a van."},

"1123244": {"semantic": [{"operation": "select", "dependencies": [], "argument": "scene"}, {"operation": "query", "dependencies": [0], "argument": "place"}], "entailed": ["1123245", "1123246", "1123247", "1123243"], "equivalent": ["1123244"], "question": "Which place is it?", "imageId": "2266", "isBalanced": true, "groups": {"global": "place", "local": "02q-place"}, "answer": "store", "semanticStr": "select: scene->query: place [0]", "annotations": {"answer": {}, "question": {}, "fullAnswer": {}}, "types": {"detailed": "place", "semantic": "global", "structural": "query"}, "fullAnswer": "It is a store."},

"1123467": {"semantic": [{"operation": "select", "dependencies": [], "argument": "people (4324184)"}, {"operation": "relate", "dependencies": [0], "argument": "man,to the right of,s (4324193)"}, {"operation": "relate", "dependencies": [1], "argument": "vehicle,to the right of,s (4537536)"}, {"operation": "query", "dependencies": [2], "argument": "name"}], "entailed": ["1123468", "1123469", "1123365"], "equivalent": ["1123467"], "question": "What is the vehicle to the right of the man that is to the right of the people?", "imageId": "2266", "isBalanced": true, "groups": {"global": "vehicle", "local": "15-man_to the right of,s"}, "answer": "car", "semanticStr": "select: people (4324184)->relate: man,to the right of,s (4324193) [0]->relate: vehicle,to the right of,s (4537536) [1]->query: name [2]", "annotations": {"answer": {"0": "4537536"}, "question": {"9": "4324193", "3": "4537536", "17": "4324184"}, "fullAnswer": {"1": "4537536", "4": "4537536"}}, "types": {"detailed": "categoryRelS", "semantic": "rel", "structural": "query"}, "fullAnswer": "The vehicle is a car."},

"1123458": {"semantic": [{"operation": "select", "dependencies": [], "argument": "car (4324192)"}, {"operation": "relate", "dependencies": [0], "argument": "car,to the right of,s (4537536)"}, {"operation": "choose color", "dependencies": [1], "argument": "black|gold"}], "entailed": ["1123457"], "equivalent": ["1123458", "1123457"], "question": "Is the car that is to the right of the other car gold or black?", "imageId": "2266", "isBalanced": true, "groups": {"global": "color", "local": "10c-car_color"}, "answer": "black", "semanticStr": "select: car (4324192)->relate: car,to the right of,s (4537536) [0]->choose color: black|gold [1]", "annotations": {"answer": {}, "question": {"11": "4324192", "2": "4537536"}, "fullAnswer": {"1": "4537536"}}, "types": {"detailed": "chooseAttr", "semantic": "attr", "structural": "choose"}, "fullAnswer": "The car is black."},

"1123259": {"semantic": [{"operation": "select", "dependencies": [], "argument": "van (4537539)"}, {"operation": "select", "dependencies": [], "argument": "shirt (4324191) "}, {"operation": "different color", "dependencies": [0, 1], "argument": ""}], "entailed": ["1123257", "1123256", "1123260", "1123261", "1123258"], "equivalent": ["1123260", "1123261", "1123258", "1123259"], "question": "Does the van have a different color than the shirt?", "imageId": "2266", "isBalanced": true, "groups": {"global": null, "local": "09diff-shirt_van"}, "answer": "no", "semanticStr": "select: van (4537539)->select: shirt (4324191) ->different color:  [0, 1]", "annotations": {"answer": {}, "question": {"9": "4324191", "2": "4537539"}, "fullAnswer": {"3": "4537539", "6": "4324191"}}, "types": {"detailed": "twoDifferentC", "semantic": "attr", "structural": "compare"}, "fullAnswer": "No, both the van and the shirt are white."},

"1123415": {"semantic": [{"operation": "select", "dependencies": [], "argument": "van (4537539)"}, {"operation": "query", "dependencies": [0], "argument": "hposition"}], "entailed": ["1123414", "1123407", "1123416", "1123411", "1123410", "1123413", "1123412", "1123408", "1123409"], "equivalent": ["1123415"], "question": "On which side of the picture is the van?", "imageId": "2266", "isBalanced": true, "groups": {"global": null, "local": "10q-van_hposition"}, "answer": "right", "semanticStr": "select: van (4537539)->query: hposition [0]", "annotations": {"answer": {}, "question": {"8": "4537539"}, "fullAnswer": {"1": "4537539"}}, "types": {"detailed": "positionQuery", "semantic": "attr", "structural": "query"}, "fullAnswer": "The van is on the right of the image."},

"1123432": {"semantic": [{"operation": "select", "dependencies": [], "argument": "street (4537538)"}, {"operation": "relate", "dependencies": [0], "argument": "vehicle,on,s (4537539)"}, {"operation": "query", "dependencies": [1], "argument": "name"}], "entailed": ["1123431", "1123436"], "equivalent": ["1123432"], "question": "What vehicle is on the street?", "imageId": "2266", "isBalanced": true, "groups": {"global": "vehicle", "local": "15-street_on,s"}, "answer": "van", "semanticStr": "select: street (4537538)->relate: vehicle,on,s (4537539) [0]->query: name [1]", "annotations": {"answer": {"0": "4537539"}, "question": {"1": "4537539"}, "fullAnswer": {"1": "4537539", "4": "4537539"}}, "types": {"detailed": "categoryRelS", "semantic": "rel", "structural": "query"}, "fullAnswer": "The vehicle is a van."},

"1123431": {"semantic": [{"operation": "select", "dependencies": [], "argument": "street (4537538)"}, {"operation": "relate", "dependencies": [0], "argument": "vehicle,on,s (4537539)"}, {"operation": "query", "dependencies": [1], "argument": "name"}], "entailed": ["1123432", "1123436"], "equivalent": ["1123431"], "question": "What is the vehicle on the street?", "imageId": "2266", "isBalanced": true, "groups": {"global": "vehicle", "local": "15-street_on,s"}, "answer": "van", "semanticStr": "select: street (4537538)->relate: vehicle,on,s (4537539) [0]->query: name [1]", "annotations": {"answer": {"0": "4537539"}, "question": {"3": "4537539", "6": "4537538"}, "fullAnswer": {"1": "4537539", "4": "4537539"}}, "types": {"detailed": "categoryRelS", "semantic": "rel", "structural": "query"}, "fullAnswer": "The vehicle is a van."},