File size: 8,634 Bytes
2ada650
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
"""
 Copyright (c) 2022, salesforce.com, inc.
 All rights reserved.
 SPDX-License-Identifier: BSD-3-Clause
 For full license text, see the LICENSE file in the repo root or https://opensource.org/licenses/BSD-3-Clause
"""

__author__ = "aagrawal"
__version__ = "0.9"

# Interface for accessing the VQA dataset.

# This code is based on the code written by Tsung-Yi Lin for MSCOCO Python API available at the following link:
# (https://github.com/pdollar/coco/blob/master/PythonAPI/pycocotools/coco.py).

# The following functions are defined:
#  VQA        - VQA class that loads VQA annotation file and prepares data structures.
#  getQuesIds - Get question ids that satisfy given filter conditions.
#  getImgIds  - Get image ids that satisfy given filter conditions.
#  loadQA     - Load questions and answers with the specified question ids.
#  showQA     - Display the specified questions and answers.
#  loadRes    - Load result file and create result object.

# Help on each function can be accessed by: "help(COCO.function)"

import json
import datetime
import copy


class VQA:
    def __init__(self, annotation_file=None, question_file=None):
        """
        Constructor of VQA helper class for reading and visualizing questions and answers.
        :param annotation_file (str): location of VQA annotation file
        :return:
        """
        # load dataset
        self.dataset = {}
        self.questions = {}
        self.qa = {}
        self.qqa = {}
        self.imgToQA = {}
        if not annotation_file == None and not question_file == None:
            print("loading VQA annotations and questions into memory...")
            time_t = datetime.datetime.utcnow()
            dataset = json.load(open(annotation_file, "r"))
            questions = json.load(open(question_file, "r"))
            self.dataset = dataset
            self.questions = questions
            self.createIndex()

    def createIndex(self):
        # create index
        print("creating index...")
        imgToQA = {ann["image_id"]: [] for ann in self.dataset["annotations"]}
        qa = {ann["question_id"]: [] for ann in self.dataset["annotations"]}
        qqa = {ann["question_id"]: [] for ann in self.dataset["annotations"]}
        for ann in self.dataset["annotations"]:
            imgToQA[ann["image_id"]] += [ann]
            qa[ann["question_id"]] = ann
        for ques in self.questions["questions"]:
            qqa[ques["question_id"]] = ques
        print("index created!")

        # create class members
        self.qa = qa
        self.qqa = qqa
        self.imgToQA = imgToQA

    def info(self):
        """
        Print information about the VQA annotation file.
        :return:
        """
        for key, value in self.datset["info"].items():
            print("%s: %s" % (key, value))

    def getQuesIds(self, imgIds=[], quesTypes=[], ansTypes=[]):
        """
        Get question ids that satisfy given filter conditions. default skips that filter
        :param 	imgIds    (int array)   : get question ids for given imgs
                        quesTypes (str array)   : get question ids for given question types
                        ansTypes  (str array)   : get question ids for given answer types
        :return:    ids   (int array)   : integer array of question ids
        """
        imgIds = imgIds if type(imgIds) == list else [imgIds]
        quesTypes = quesTypes if type(quesTypes) == list else [quesTypes]
        ansTypes = ansTypes if type(ansTypes) == list else [ansTypes]

        if len(imgIds) == len(quesTypes) == len(ansTypes) == 0:
            anns = self.dataset["annotations"]
        else:
            if not len(imgIds) == 0:
                anns = sum(
                    [self.imgToQA[imgId] for imgId in imgIds if imgId in self.imgToQA],
                    [],
                )
            else:
                anns = self.dataset["annotations"]
            anns = (
                anns
                if len(quesTypes) == 0
                else [ann for ann in anns if ann["question_type"] in quesTypes]
            )
            anns = (
                anns
                if len(ansTypes) == 0
                else [ann for ann in anns if ann["answer_type"] in ansTypes]
            )
        ids = [ann["question_id"] for ann in anns]
        return ids

    def getImgIds(self, quesIds=[], quesTypes=[], ansTypes=[]):
        """
         Get image ids that satisfy given filter conditions. default skips that filter
         :param quesIds   (int array)   : get image ids for given question ids
        quesTypes (str array)   : get image ids for given question types
        ansTypes  (str array)   : get image ids for given answer types
         :return: ids     (int array)   : integer array of image ids
        """
        quesIds = quesIds if type(quesIds) == list else [quesIds]
        quesTypes = quesTypes if type(quesTypes) == list else [quesTypes]
        ansTypes = ansTypes if type(ansTypes) == list else [ansTypes]

        if len(quesIds) == len(quesTypes) == len(ansTypes) == 0:
            anns = self.dataset["annotations"]
        else:
            if not len(quesIds) == 0:
                anns = sum(
                    [self.qa[quesId] for quesId in quesIds if quesId in self.qa], []
                )
            else:
                anns = self.dataset["annotations"]
            anns = (
                anns
                if len(quesTypes) == 0
                else [ann for ann in anns if ann["question_type"] in quesTypes]
            )
            anns = (
                anns
                if len(ansTypes) == 0
                else [ann for ann in anns if ann["answer_type"] in ansTypes]
            )
        ids = [ann["image_id"] for ann in anns]
        return ids

    def loadQA(self, ids=[]):
        """
        Load questions and answers with the specified question ids.
        :param ids (int array)       : integer ids specifying question ids
        :return: qa (object array)   : loaded qa objects
        """
        if type(ids) == list:
            return [self.qa[id] for id in ids]
        elif type(ids) == int:
            return [self.qa[ids]]

    def showQA(self, anns):
        """
        Display the specified annotations.
        :param anns (array of object): annotations to display
        :return: None
        """
        if len(anns) == 0:
            return 0
        for ann in anns:
            quesId = ann["question_id"]
            print("Question: %s" % (self.qqa[quesId]["question"]))
            for ans in ann["answers"]:
                print("Answer %d: %s" % (ans["answer_id"], ans["answer"]))

    def loadRes(self, resFile, quesFile):
        """
        Load result file and return a result object.
        :param   resFile (str)     : file name of result file
        :return: res (obj)         : result api object
        """
        res = VQA()
        res.questions = json.load(open(quesFile))
        res.dataset["info"] = copy.deepcopy(self.questions["info"])
        res.dataset["task_type"] = copy.deepcopy(self.questions["task_type"])
        res.dataset["data_type"] = copy.deepcopy(self.questions["data_type"])
        res.dataset["data_subtype"] = copy.deepcopy(self.questions["data_subtype"])
        res.dataset["license"] = copy.deepcopy(self.questions["license"])

        print("Loading and preparing results...     ")
        time_t = datetime.datetime.utcnow()
        anns = json.load(open(resFile))
        assert type(anns) == list, "results is not an array of objects"
        annsQuesIds = [ann["question_id"] for ann in anns]
        assert set(annsQuesIds) == set(
            self.getQuesIds()
        ), "Results do not correspond to current VQA set. Either the results do not have predictions for all question ids in annotation file or there is atleast one question id that does not belong to the question ids in the annotation file."
        for ann in anns:
            quesId = ann["question_id"]
            if res.dataset["task_type"] == "Multiple Choice":
                assert (
                    ann["answer"] in self.qqa[quesId]["multiple_choices"]
                ), "predicted answer is not one of the multiple choices"
            qaAnn = self.qa[quesId]
            ann["image_id"] = qaAnn["image_id"]
            ann["question_type"] = qaAnn["question_type"]
            ann["answer_type"] = qaAnn["answer_type"]
        print(
            "DONE (t=%0.2fs)" % ((datetime.datetime.utcnow() - time_t).total_seconds())
        )

        res.dataset["annotations"] = anns
        res.createIndex()
        return res