Duplicate from ranjaykrishna/visual_genome

Co-authored-by: Parquet-converter (BOT) <parquet-converter@users.noreply.huggingface.co>

.gitattributes

@@ -0,0 +1,38 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bin.* filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zstandard filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+# Audio files - uncompressed
+*.pcm filter=lfs diff=lfs merge=lfs -text
+*.sam filter=lfs diff=lfs merge=lfs -text
+*.raw filter=lfs diff=lfs merge=lfs -text
+# Audio files - compressed
+*.aac filter=lfs diff=lfs merge=lfs -text
+*.flac filter=lfs diff=lfs merge=lfs -text
+*.mp3 filter=lfs diff=lfs merge=lfs -text
+*.ogg filter=lfs diff=lfs merge=lfs -text
+*.wav filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,503 @@
+---
+annotations_creators:
+- found
+language_creators:
+- found
+language:
+- en
+license:
+- cc-by-4.0
+multilinguality:
+- monolingual
+size_categories:
+- 100K<n<1M
+source_datasets:
+- original
+task_categories:
+- image-to-text
+- object-detection
+- visual-question-answering
+task_ids:
+- image-captioning
+paperswithcode_id: visual-genome
+pretty_name: VisualGenome
+dataset_info:
+  features:
+  - name: image
+    dtype: image
+  - name: image_id
+    dtype: int32
+  - name: url
+    dtype: string
+  - name: width
+    dtype: int32
+  - name: height
+    dtype: int32
+  - name: coco_id
+    dtype: int64
+  - name: flickr_id
+    dtype: int64
+  - name: regions
+    list:
+    - name: region_id
+      dtype: int32
+    - name: image_id
+      dtype: int32
+    - name: phrase
+      dtype: string
+    - name: x
+      dtype: int32
+    - name: y
+      dtype: int32
+    - name: width
+      dtype: int32
+    - name: height
+      dtype: int32
+  config_name: region_descriptions_v1.0.0
+  splits:
+  - name: train
+    num_bytes: 260873884
+    num_examples: 108077
+  download_size: 15304605295
+  dataset_size: 260873884
+config_names:
+- objects
+- question_answers
+- region_descriptions
+---
+
+# Dataset Card for Visual Genome
+
+## Table of Contents
+- [Table of Contents](#table-of-contents)
+- [Dataset Description](#dataset-description)
+  - [Dataset Summary](#dataset-summary)
+  - [Dataset Preprocessing](#dataset-preprocessing)
+  - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
+  - [Languages](#languages)
+- [Dataset Structure](#dataset-structure)
+  - [Data Instances](#data-instances)
+  - [Data Fields](#data-fields)
+  - [Data Splits](#data-splits)
+- [Dataset Creation](#dataset-creation)
+  - [Curation Rationale](#curation-rationale)
+  - [Source Data](#source-data)
+  - [Annotations](#annotations)
+  - [Personal and Sensitive Information](#personal-and-sensitive-information)
+- [Considerations for Using the Data](#considerations-for-using-the-data)
+  - [Social Impact of Dataset](#social-impact-of-dataset)
+  - [Discussion of Biases](#discussion-of-biases)
+  - [Other Known Limitations](#other-known-limitations)
+- [Additional Information](#additional-information)
+  - [Dataset Curators](#dataset-curators)
+  - [Licensing Information](#licensing-information)
+  - [Citation Information](#citation-information)
+  - [Contributions](#contributions)
+
+## Dataset Description
+
+- **Homepage:** https://homes.cs.washington.edu/~ranjay/visualgenome/
+- **Repository:** 
+- **Paper:** https://doi.org/10.1007/s11263-016-0981-7
+- **Leaderboard:**
+- **Point of Contact:** ranjaykrishna [at] gmail [dot] com
+
+### Dataset Summary
+
+Visual Genome is a dataset, a knowledge base, an ongoing effort to connect structured image concepts to language.
+
+From the paper:
+> Despite progress in perceptual tasks such as
+image classification, computers still perform poorly on
+cognitive tasks such as image description and question
+answering. Cognition is core to tasks that involve not
+just recognizing, but reasoning about our visual world.
+However, models used to tackle the rich content in images for cognitive tasks are still being trained using the
+same datasets designed for perceptual tasks. To achieve
+success at cognitive tasks, models need to understand
+the interactions and relationships between objects in an
+image. When asked “What vehicle is the person riding?”,
+computers will need to identify the objects in an image
+as well as the relationships riding(man, carriage) and
+pulling(horse, carriage) to answer correctly that “the
+person is riding a horse-drawn carriage.”
+
+Visual Genome has:
+ - 108,077 image
+ - 5.4 Million Region Descriptions
+ - 1.7 Million Visual Question Answers
+ - 3.8 Million Object Instances
+ - 2.8 Million Attributes
+ - 2.3 Million Relationships
+
+From the paper:
+> Our dataset contains over 108K images where each
+image has an average of 35 objects, 26 attributes, and 21
+pairwise relationships between objects. We canonicalize
+the objects, attributes, relationships, and noun phrases
+in region descriptions and questions answer pairs to
+WordNet synsets.
+
+### Dataset Preprocessing
+
+### Supported Tasks and Leaderboards
+
+### Languages
+
+All of annotations use English as primary language.
+
+## Dataset Structure
+
+### Data Instances
+
+When loading a specific configuration, users has to append a version dependent suffix:
+```python
+from datasets import load_dataset
+load_dataset("visual_genome", "region_description_v1.2.0")
+```
+
+#### region_descriptions
+
+An example of looks as follows.
+
+```
+{
+  "image": <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=800x600 at 0x7F2F60698610>,
+  "image_id": 1,
+  "url": "https://cs.stanford.edu/people/rak248/VG_100K_2/1.jpg",
+  "width": 800,
+  "height": 600,
+  "coco_id": null,
+  "flickr_id": null,
+  "regions": [
+    {
+      "region_id": 1382,
+      "image_id": 1,
+      "phrase": "the clock is green in colour",
+      "x": 421,
+      "y": 57,
+      "width": 82,
+      "height": 139
+    },
+    ...
+  ]
+}
+```
+
+#### objects
+
+An example of looks as follows.
+
+```
+{
+  "image": <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=800x600 at 0x7F2F60698610>,
+  "image_id": 1,
+  "url": "https://cs.stanford.edu/people/rak248/VG_100K_2/1.jpg",
+  "width": 800,
+  "height": 600,
+  "coco_id": null,
+  "flickr_id": null,
+  "objects": [
+    {
+      "object_id": 1058498,
+      "x": 421,
+      "y": 91,
+      "w": 79,
+      "h": 339,
+      "names": [
+        "clock"
+      ],
+      "synsets": [
+        "clock.n.01"
+      ]
+    },
+    ...
+  ]
+}
+```
+
+#### attributes
+
+An example of looks as follows.
+
+```
+{
+  "image": <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=800x600 at 0x7F2F60698610>,
+  "image_id": 1,
+  "url": "https://cs.stanford.edu/people/rak248/VG_100K_2/1.jpg",
+  "width": 800,
+  "height": 600,
+  "coco_id": null,
+  "flickr_id": null,
+  "attributes": [
+    {
+      "object_id": 1058498,
+      "x": 421,
+      "y": 91,
+      "w": 79,
+      "h": 339,
+      "names": [
+        "clock"
+      ],
+      "synsets": [
+        "clock.n.01"
+      ],
+      "attributes": [
+        "green",
+        "tall"
+      ]
+    },
+    ...
+  }
+]
+```
+
+#### relationships
+
+An example of looks as follows.
+
+```
+{
+  "image": <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=800x600 at 0x7F2F60698610>,
+  "image_id": 1,
+  "url": "https://cs.stanford.edu/people/rak248/VG_100K_2/1.jpg",
+  "width": 800,
+  "height": 600,
+  "coco_id": null,
+  "flickr_id": null,
+  "relationships": [
+    {
+      "relationship_id": 15927,
+      "predicate": "ON",
+      "synsets": "['along.r.01']",
+      "subject": {
+        "object_id": 5045,
+        "x": 119,
+        "y": 338,
+        "w": 274,
+        "h": 192,
+        "names": [
+          "shade"
+        ],
+        "synsets": [
+          "shade.n.01"
+        ]
+      },
+      "object": {
+        "object_id": 5046,
+        "x": 77,
+        "y": 328,
+        "w": 714,
+        "h": 262,
+        "names": [
+          "street"
+        ],
+        "synsets": [
+          "street.n.01"
+        ]
+      }
+    }
+    ...
+  }
+]
+```
+#### question_answers
+
+An example of looks as follows.
+
+```
+{
+  "image": <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=800x600 at 0x7F2F60698610>,
+  "image_id": 1,
+  "url": "https://cs.stanford.edu/people/rak248/VG_100K_2/1.jpg",
+  "width": 800,
+  "height": 600,
+  "coco_id": null,
+  "flickr_id": null,
+  "qas": [
+    {
+      "qa_id": 986768,
+      "image_id": 1,
+      "question": "What color is the clock?",
+      "answer": "Green.",
+      "a_objects": [],
+      "q_objects": []
+    },
+    ...
+  }
+]
+```
+
+### Data Fields
+
+When loading a specific configuration, users has to append a version dependent suffix:
+```python
+from datasets import load_dataset
+load_dataset("visual_genome", "region_description_v1.2.0")
+```
+
+#### region_descriptions
+
+- `image`: A `PIL.Image.Image` object containing the image. Note that when accessing the image column: `dataset[0]["image"]` the image file is automatically decoded. Decoding of a large number of image files might take a significant amount of time. Thus it is important to first query the sample index before the `"image"` column, *i.e.* `dataset[0]["image"]` should **always** be preferred over `dataset["image"][0]`
+- `image_id`: Unique numeric ID of the image.
+- `url`:  URL of source image.
+- `width`: Image width.
+- `height`: Image height.
+- `coco_id`: Id mapping to MSCOCO indexing.
+- `flickr_id`: Id mapping to Flicker indexing.
+- `regions`: Holds a list of `Region` dataclasses:
+  - `region_id`: Unique numeric ID of the region.
+  - `image_id`: Unique numeric ID of the image.
+  - `x`: x coordinate of bounding box's top left corner.
+  - `y`: y coordinate of bounding box's top left corner.
+  - `width`: Bounding box width.
+  - `height`: Bounding box height.
+
+#### objects
+
+- `image`: A `PIL.Image.Image` object containing the image. Note that when accessing the image column: `dataset[0]["image"]` the image file is automatically decoded. Decoding of a large number of image files might take a significant amount of time. Thus it is important to first query the sample index before the `"image"` column, *i.e.* `dataset[0]["image"]` should **always** be preferred over `dataset["image"][0]`
+- `image_id`: Unique numeric ID of the image.
+- `url`:  URL of source image.
+- `width`: Image width.
+- `height`: Image height.
+- `coco_id`: Id mapping to MSCOCO indexing.
+- `flickr_id`: Id mapping to Flicker indexing.
+- `objects`: Holds a list of `Object` dataclasses:
+  - `object_id`: Unique numeric ID of the object.
+  - `x`: x coordinate of bounding box's top left corner.
+  - `y`: y coordinate of bounding box's top left corner.
+  - `w`: Bounding box width.
+  - `h`: Bounding box height.
+  - `names`: List of names associated with the object. This field can hold multiple values in the sense the multiple names are considered as acceptable. For example: ['monitor', 'computer'] at https://cs.stanford.edu/people/rak248/VG_100K/3.jpg
+  - `synsets`: List of `WordNet synsets`.
+
+#### attributes
+
+- `image`: A `PIL.Image.Image` object containing the image. Note that when accessing the image column: `dataset[0]["image"]` the image file is automatically decoded. Decoding of a large number of image files might take a significant amount of time. Thus it is important to first query the sample index before the `"image"` column, *i.e.* `dataset[0]["image"]` should **always** be preferred over `dataset["image"][0]`
+- `image_id`: Unique numeric ID of the image.
+- `url`:  URL of source image.
+- `width`: Image width.
+- `height`: Image height.
+- `coco_id`: Id mapping to MSCOCO indexing.
+- `flickr_id`: Id mapping to Flicker indexing.
+- `attributes`: Holds a list of `Object` dataclasses:
+  - `object_id`: Unique numeric ID of the region.
+  - `x`: x coordinate of bounding box's top left corner.
+  - `y`: y coordinate of bounding box's top left corner.
+  - `w`: Bounding box width.
+  - `h`: Bounding box height.
+  - `names`: List of names associated with the object. This field can hold multiple values in the sense the multiple names are considered as acceptable. For example: ['monitor', 'computer'] at https://cs.stanford.edu/people/rak248/VG_100K/3.jpg
+  - `synsets`: List of `WordNet synsets`.
+  - `attributes`: List of attributes associated with the object.
+
+#### relationships
+
+- `image`: A `PIL.Image.Image` object containing the image. Note that when accessing the image column: `dataset[0]["image"]` the image file is automatically decoded. Decoding of a large number of image files might take a significant amount of time. Thus it is important to first query the sample index before the `"image"` column, *i.e.* `dataset[0]["image"]` should **always** be preferred over `dataset["image"][0]`
+- `image_id`: Unique numeric ID of the image.
+- `url`:  URL of source image.
+- `width`: Image width.
+- `height`: Image height.
+- `coco_id`: Id mapping to MSCOCO indexing.
+- `flickr_id`: Id mapping to Flicker indexing.
+- `relationships`: Holds a list of `Relationship` dataclasses:
+  - `relationship_id`: Unique numeric ID of the object.
+  - `predicate`: Predicate defining relationship between a subject and an object.
+  - `synsets`: List of `WordNet synsets`.
+  - `subject`: Object dataclass. See subsection on `objects`.
+  - `object`: Object dataclass. See subsection on `objects`.
+
+#### question_answers
+
+- `image`: A `PIL.Image.Image` object containing the image. Note that when accessing the image column: `dataset[0]["image"]` the image file is automatically decoded. Decoding of a large number of image files might take a significant amount of time. Thus it is important to first query the sample index before the `"image"` column, *i.e.* `dataset[0]["image"]` should **always** be preferred over `dataset["image"][0]`
+- `image_id`: Unique numeric ID of the image.
+- `url`:  URL of source image.
+- `width`: Image width.
+- `height`: Image height.
+- `coco_id`: Id mapping to MSCOCO indexing.
+- `flickr_id`: Id mapping to Flicker indexing.
+- `qas`: Holds a list of `Question-Answering` dataclasses:
+  - `qa_id`: Unique numeric ID of the question-answer pair.
+  - `image_id`: Unique numeric ID of the image.
+  - `question`: Question.
+  - `answer`: Answer.
+  - `q_objects`: List of object dataclass associated with `question` field. See subsection on `objects`.
+  - `a_objects`: List of object dataclass associated with `answer` field. See subsection on `objects`.
+
+### Data Splits
+
+All the data is contained in training set.
+
+## Dataset Creation
+
+### Curation Rationale
+
+### Source Data
+
+#### Initial Data Collection and Normalization
+
+#### Who are the source language producers?
+
+### Annotations
+
+#### Annotation process
+
+#### Who are the annotators?
+
+From the paper:
+> We used Amazon Mechanical Turk (AMT) as our primary source of annotations. Overall, a total of over
+33, 000 unique workers contributed to the dataset. The
+dataset was collected over the course of 6 months after
+15 months of experimentation and iteration on the data
+representation. Approximately 800, 000 Human Intelligence Tasks (HITs) were launched on AMT, where
+each HIT involved creating descriptions, questions and
+answers, or region graphs. Each HIT was designed such
+that workers manage to earn anywhere between $6-$8
+per hour if they work continuously, in line with ethical
+research standards on Mechanical Turk (Salehi et al.,
+2015). Visual Genome HITs achieved a 94.1% retention
+rate, meaning that 94.1% of workers who completed one
+of our tasks went ahead to do more. [...] 93.02% of workers contributed from the United States.
+The majority of our workers were
+between the ages of 25 and 34 years old. Our youngest
+contributor was 18 years and the oldest was 68 years
+old. We also had a near-balanced split of 54.15% male
+and 45.85% female workers.
+
+### Personal and Sensitive Information
+
+## Considerations for Using the Data
+
+### Social Impact of Dataset
+
+### Discussion of Biases
+
+### Other Known Limitations
+
+## Additional Information
+
+### Dataset Curators
+
+### Licensing Information
+
+Visual Genome by Ranjay Krishna is licensed under a Creative Commons Attribution 4.0 International License.
+
+### Citation Information
+
+```bibtex
+@article{Krishna2016VisualGC,
+  title={Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotations},
+  author={Ranjay Krishna and Yuke Zhu and Oliver Groth and Justin Johnson and Kenji Hata and Joshua Kravitz and Stephanie Chen and Yannis Kalantidis and Li-Jia Li and David A. Shamma and Michael S. Bernstein and Li Fei-Fei},
+  journal={International Journal of Computer Vision},
+  year={2017},
+  volume={123},
+  pages={32-73},
+  url={https://doi.org/10.1007/s11263-016-0981-7},
+  doi={10.1007/s11263-016-0981-7}
+}
+```
+
+### Contributions
+
+Due to limitation of the dummy_data creation, we provide a `fix_generated_dummy_data.py` script that fix the dataset in-place.
+
+Thanks to [@thomasw21](https://github.com/thomasw21) for adding this dataset.

visual_genome.py

@@ -0,0 +1,469 @@
+# coding=utf-8
+# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""Visual Genome dataset."""
+
+import json
+import os
+import re
+from collections import defaultdict
+from typing import Any, Callable, Dict, Optional
+from urllib.parse import urlparse
+
+import datasets
+
+
+logger = datasets.logging.get_logger(__name__)
+
+_CITATION = """\
+@article{Krishna2016VisualGC,
+  title={Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotations},
+  author={Ranjay Krishna and Yuke Zhu and Oliver Groth and Justin Johnson and Kenji Hata and Joshua Kravitz and Stephanie Chen and Yannis Kalantidis and Li-Jia Li and David A. Shamma and Michael S. Bernstein and Li Fei-Fei},
+  journal={International Journal of Computer Vision},
+  year={2017},
+  volume={123},
+  pages={32-73},
+  url={https://doi.org/10.1007/s11263-016-0981-7},
+  doi={10.1007/s11263-016-0981-7}
+}
+"""
+
+_DESCRIPTION = """\
+Visual Genome enable to model objects and relationships between objects.
+They collect dense annotations of objects, attributes, and relationships within each image.
+Specifically, the dataset contains over 108K images where each image has an average of 35 objects, 26 attributes, and 21 pairwise relationships between objects.
+"""
+
+_HOMEPAGE = "https://homes.cs.washington.edu/~ranjay/visualgenome/"
+
+_LICENSE = "Creative Commons Attribution 4.0 International License"
+
+_BASE_IMAGE_URLS = {
+    "https://cs.stanford.edu/people/rak248/VG_100K_2/images.zip": "VG_100K",
+    "https://cs.stanford.edu/people/rak248/VG_100K_2/images2.zip": "VG_100K_2",
+}
+
+_LATEST_VERSIONS = {
+    "region_descriptions": "1.2.0",
+    "objects": "1.4.0",
+    "attributes": "1.2.0",
+    "relationships": "1.4.0",
+    "question_answers": "1.2.0",
+    "image_metadata": "1.2.0",
+}
+
+# ---- Features ----
+
+_BASE_IMAGE_METADATA_FEATURES = {
+    "image_id": datasets.Value("int32"),
+    "url": datasets.Value("string"),
+    "width": datasets.Value("int32"),
+    "height": datasets.Value("int32"),
+    "coco_id": datasets.Value("int64"),
+    "flickr_id": datasets.Value("int64"),
+}
+
+_BASE_SYNTET_FEATURES = {
+    "synset_name": datasets.Value("string"),
+    "entity_name": datasets.Value("string"),
+    "entity_idx_start": datasets.Value("int32"),
+    "entity_idx_end": datasets.Value("int32"),
+}
+
+_BASE_OBJECT_FEATURES = {
+    "object_id": datasets.Value("int32"),
+    "x": datasets.Value("int32"),
+    "y": datasets.Value("int32"),
+    "w": datasets.Value("int32"),
+    "h": datasets.Value("int32"),
+    "names": [datasets.Value("string")],
+    "synsets": [datasets.Value("string")],
+}
+
+_BASE_QA_OBJECT_FEATURES = {
+    "object_id": datasets.Value("int32"),
+    "x": datasets.Value("int32"),
+    "y": datasets.Value("int32"),
+    "w": datasets.Value("int32"),
+    "h": datasets.Value("int32"),
+    "names": [datasets.Value("string")],
+    "synsets": [datasets.Value("string")],
+}
+
+_BASE_QA_OBJECT = {
+    "qa_id": datasets.Value("int32"),
+    "image_id": datasets.Value("int32"),
+    "question": datasets.Value("string"),
+    "answer": datasets.Value("string"),
+    "a_objects": [_BASE_QA_OBJECT_FEATURES],
+    "q_objects": [_BASE_QA_OBJECT_FEATURES],
+}
+
+_BASE_REGION_FEATURES = {
+    "region_id": datasets.Value("int32"),
+    "image_id": datasets.Value("int32"),
+    "phrase": datasets.Value("string"),
+    "x": datasets.Value("int32"),
+    "y": datasets.Value("int32"),
+    "width": datasets.Value("int32"),
+    "height": datasets.Value("int32"),
+}
+
+_BASE_RELATIONSHIP_FEATURES = {
+    "relationship_id": datasets.Value("int32"),
+    "predicate": datasets.Value("string"),
+    "synsets": datasets.Value("string"),
+    "subject": _BASE_OBJECT_FEATURES,
+    "object": _BASE_OBJECT_FEATURES,
+}
+
+_NAME_VERSION_TO_ANNOTATION_FEATURES = {
+    "region_descriptions": {
+        "1.2.0": {"regions": [_BASE_REGION_FEATURES]},
+        "1.0.0": {"regions": [_BASE_REGION_FEATURES]},
+    },
+    "objects": {
+        "1.4.0": {"objects": [{**_BASE_OBJECT_FEATURES, "merged_object_ids": [datasets.Value("int32")]}]},
+        "1.2.0": {"objects": [_BASE_OBJECT_FEATURES]},
+        "1.0.0": {"objects": [_BASE_OBJECT_FEATURES]},
+    },
+    "attributes": {
+        "1.2.0": {"attributes": [{**_BASE_OBJECT_FEATURES, "attributes": [datasets.Value("string")]}]},
+        "1.0.0": {"attributes": [{**_BASE_OBJECT_FEATURES, "attributes": [datasets.Value("string")]}]},
+    },
+    "relationships": {
+        "1.4.0": {
+            "relationships": [
+                {
+                    **_BASE_RELATIONSHIP_FEATURES,
+                    "subject": {**_BASE_OBJECT_FEATURES, "merged_object_ids": [datasets.Value("int32")]},
+                    "object": {**_BASE_OBJECT_FEATURES, "merged_object_ids": [datasets.Value("int32")]},
+                }
+            ]
+        },
+        "1.2.0": {"relationships": [_BASE_RELATIONSHIP_FEATURES]},
+        "1.0.0": {"relationships": [_BASE_RELATIONSHIP_FEATURES]},
+    },
+    "question_answers": {"1.2.0": {"qas": [_BASE_QA_OBJECT]}, "1.0.0": {"qas": [_BASE_QA_OBJECT]}},
+}
+
+# ----- Helpers -----
+
+
+def _get_decompressed_filename_from_url(url: str) -> str:
+    parsed_url = urlparse(url)
+    compressed_filename = os.path.basename(parsed_url.path)
+
+    # Remove `.zip` suffix
+    assert compressed_filename.endswith(".zip")
+    uncompressed_filename = compressed_filename[:-4]
+
+    # Remove version.
+    unversioned_uncompressed_filename = re.sub(r"_v[0-9]+(?:_[0-9]+)?\.json$", ".json", uncompressed_filename)
+
+    return unversioned_uncompressed_filename
+
+
+def _get_local_image_path(img_url: str, folder_local_paths: Dict[str, str]) -> str:
+    """
+    Obtain image folder given an image url.
+
+    For example:
+      Given `https://cs.stanford.edu/people/rak248/VG_100K_2/1.jpg` as an image url, this method returns the local path for that image.
+    """
+    matches = re.fullmatch(r"^https://cs.stanford.edu/people/rak248/(VG_100K(?:_2)?)/([0-9]+\.jpg)$", img_url)
+    assert matches is not None, f"Got img_url: {img_url}, matched: {matches}"
+    folder, filename = matches.group(1), matches.group(2)
+    return os.path.join(folder_local_paths[folder], filename)
+
+
+# ----- Annotation normalizers ----
+
+_BASE_ANNOTATION_URL = "https://homes.cs.washington.edu/~ranjay/visualgenome/data/dataset"
+
+
+def _normalize_region_description_annotation_(annotation: Dict[str, Any]) -> Dict[str, Any]:
+    """Normalizes region descriptions annotation in-place"""
+    # Some attributes annotations don't have an attribute field
+    for region in annotation["regions"]:
+        # `id` should be converted to `region_id`:
+        if "id" in region:
+            region["region_id"] = region["id"]
+            del region["id"]
+
+        # `image` should be converted to `image_id`
+        if "image" in region:
+            region["image_id"] = region["image"]
+            del region["image"]
+
+    return annotation
+
+
+def _normalize_object_annotation_(annotation: Dict[str, Any]) -> Dict[str, Any]:
+    """Normalizes object annotation in-place"""
+    # Some attributes annotations don't have an attribute field
+    for object_ in annotation["objects"]:
+        # `id` should be converted to `object_id`:
+        if "id" in object_:
+            object_["object_id"] = object_["id"]
+            del object_["id"]
+
+        # Some versions of `object` annotations don't have `synsets` field.
+        if "synsets" not in object_:
+            object_["synsets"] = None
+
+    return annotation
+
+
+def _normalize_attribute_annotation_(annotation: Dict[str, Any]) -> Dict[str, Any]:
+    """Normalizes attributes annotation in-place"""
+    # Some attributes annotations don't have an attribute field
+    for attribute in annotation["attributes"]:
+        # `id` should be converted to `object_id`:
+        if "id" in attribute:
+            attribute["object_id"] = attribute["id"]
+            del attribute["id"]
+
+        # `objects_names` should be convered to `names:
+        if "object_names" in attribute:
+            attribute["names"] = attribute["object_names"]
+            del attribute["object_names"]
+
+        # Some versions of `attribute` annotations don't have `synsets` field.
+        if "synsets" not in attribute:
+            attribute["synsets"] = None
+
+        # Some versions of `attribute` annotations don't have `attributes` field.
+        if "attributes" not in attribute:
+            attribute["attributes"] = None
+
+    return annotation
+
+
+def _normalize_relationship_annotation_(annotation: Dict[str, Any]) -> Dict[str, Any]:
+    """Normalizes relationship annotation in-place"""
+    # For some reason relationships objects have a single name instead of a list of names.
+    for relationship in annotation["relationships"]:
+        # `id` should be converted to `object_id`:
+        if "id" in relationship:
+            relationship["relationship_id"] = relationship["id"]
+            del relationship["id"]
+
+        if "synsets" not in relationship:
+            relationship["synsets"] = None
+
+        subject = relationship["subject"]
+        object_ = relationship["object"]
+
+        for obj in [subject, object_]:
+            # `id` should be converted to `object_id`:
+            if "id" in obj:
+                obj["object_id"] = obj["id"]
+                del obj["id"]
+
+            if "name" in obj:
+                obj["names"] = [obj["name"]]
+                del obj["name"]
+
+            if "synsets" not in obj:
+                obj["synsets"] = None
+
+    return annotation
+
+
+def _normalize_image_metadata_(image_metadata: Dict[str, Any]) -> Dict[str, Any]:
+    """Normalizes image metadata in-place"""
+    if "id" in image_metadata:
+        image_metadata["image_id"] = image_metadata["id"]
+        del image_metadata["id"]
+    return image_metadata
+
+
+_ANNOTATION_NORMALIZER = defaultdict(lambda: lambda x: x)
+_ANNOTATION_NORMALIZER.update(
+    {
+        "region_descriptions": _normalize_region_description_annotation_,
+        "objects": _normalize_object_annotation_,
+        "attributes": _normalize_attribute_annotation_,
+        "relationships": _normalize_relationship_annotation_,
+    }
+)
+
+# ---- Visual Genome loading script ----
+
+
+class VisualGenomeConfig(datasets.BuilderConfig):
+    """BuilderConfig for Visual Genome."""
+
+    def __init__(self, name: str, version: Optional[str] = None, with_image: bool = True, **kwargs):
+        _version = _LATEST_VERSIONS[name] if version is None else version
+        _name = f"{name}_v{_version}"
+        super(VisualGenomeConfig, self).__init__(version=datasets.Version(_version), name=_name, **kwargs)
+        self._name_without_version = name
+        self.annotations_features = _NAME_VERSION_TO_ANNOTATION_FEATURES[self._name_without_version][
+            self.version.version_str
+        ]
+        self.with_image = with_image
+
+    @property
+    def annotations_url(self):
+        if self.version == _LATEST_VERSIONS[self._name_without_version]:
+            return f"{_BASE_ANNOTATION_URL}/{self._name_without_version}.json.zip"
+
+        major, minor = self.version.major, self.version.minor
+        if minor == 0:
+            return f"{_BASE_ANNOTATION_URL}/{self._name_without_version}_v{major}.json.zip"
+        else:
+            return f"{_BASE_ANNOTATION_URL}/{self._name_without_version}_v{major}_{minor}.json.zip"
+
+    @property
+    def image_metadata_url(self):
+        if not self.version == _LATEST_VERSIONS["image_metadata"]:
+            logger.warning(
+                f"Latest image metadata version is {_LATEST_VERSIONS['image_metadata']}. Trying to generate a dataset of version: {self.version}. Please double check that image data are unchanged between the two versions."
+            )
+        return f"{_BASE_ANNOTATION_URL}/image_data.json.zip"
+
+    @property
+    def features(self):
+        return datasets.Features(
+            {
+                **({"image": datasets.Image()} if self.with_image else {}),
+                **_BASE_IMAGE_METADATA_FEATURES,
+                **self.annotations_features,
+            }
+        )
+
+
+class VisualGenome(datasets.GeneratorBasedBuilder):
+    """Visual Genome dataset."""
+
+    BUILDER_CONFIG_CLASS = VisualGenomeConfig
+    BUILDER_CONFIGS = [
+        *[VisualGenomeConfig(name="region_descriptions", version=version) for version in ["1.0.0", "1.2.0"]],
+        *[VisualGenomeConfig(name="question_answers", version=version) for version in ["1.0.0", "1.2.0"]],
+        *[
+            VisualGenomeConfig(name="objects", version=version)
+            # TODO: add support for 1.4.0
+            for version in ["1.0.0", "1.2.0"]
+        ],
+        *[VisualGenomeConfig(name="attributes", version=version) for version in ["1.0.0", "1.2.0"]],
+        *[
+            VisualGenomeConfig(name="relationships", version=version)
+            # TODO: add support for 1.4.0
+            for version in ["1.0.0", "1.2.0"]
+        ],
+    ]
+
+    def _info(self):
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=self.config.features,
+            homepage=_HOMEPAGE,
+            license=_LICENSE,
+            citation=_CITATION,
+            version=self.config.version,
+        )
+
+    def _split_generators(self, dl_manager):
+        # Download image meta datas.
+        image_metadatas_dir = dl_manager.download_and_extract(self.config.image_metadata_url)
+        image_metadatas_file = os.path.join(
+            image_metadatas_dir, _get_decompressed_filename_from_url(self.config.image_metadata_url)
+        )
+
+        # Download annotations
+        annotations_dir = dl_manager.download_and_extract(self.config.annotations_url)
+        annotations_file = os.path.join(
+            annotations_dir, _get_decompressed_filename_from_url(self.config.annotations_url)
+        )
+
+        # Optionally download images
+        if self.config.with_image:
+            image_folder_keys = list(_BASE_IMAGE_URLS.keys())
+            image_dirs = dl_manager.download_and_extract(image_folder_keys)
+            image_folder_local_paths = {
+                _BASE_IMAGE_URLS[key]: os.path.join(dir_, _BASE_IMAGE_URLS[key])
+                for key, dir_ in zip(image_folder_keys, image_dirs)
+            }
+        else:
+            image_folder_local_paths = None
+
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TRAIN,
+                gen_kwargs={
+                    "image_folder_local_paths": image_folder_local_paths,
+                    "image_metadatas_file": image_metadatas_file,
+                    "annotations_file": annotations_file,
+                    "annotation_normalizer_": _ANNOTATION_NORMALIZER[self.config._name_without_version],
+                },
+            ),
+        ]
+
+    def _generate_examples(
+        self,
+        image_folder_local_paths: Optional[Dict[str, str]],
+        image_metadatas_file: str,
+        annotations_file: str,
+        annotation_normalizer_: Callable[[Dict[str, Any]], Dict[str, Any]],
+    ):
+        with open(annotations_file, "r", encoding="utf-8") as fi:
+            annotations = json.load(fi)
+
+        with open(image_metadatas_file, "r", encoding="utf-8") as fi:
+            image_metadatas = json.load(fi)
+
+        assert len(image_metadatas) == len(annotations)
+        for idx, (image_metadata, annotation) in enumerate(zip(image_metadatas, annotations)):
+            # in-place operation to normalize image_metadata
+            _normalize_image_metadata_(image_metadata)
+
+            # Normalize image_id across all annotations
+            if "id" in annotation:
+                # annotation["id"] corresponds to image_metadata["image_id"]
+                assert (
+                    image_metadata["image_id"] == annotation["id"]
+                ), f"Annotations doesn't match with image metadataset. Got image_metadata['image_id']: {image_metadata['image_id']} and annotations['id']: {annotation['id']}"
+                del annotation["id"]
+            else:
+                assert "image_id" in annotation
+                assert (
+                    image_metadata["image_id"] == annotation["image_id"]
+                ), f"Annotations doesn't match with image metadataset. Got image_metadata['image_id']: {image_metadata['image_id']} and annotations['image_id']: {annotation['image_id']}"
+
+            # Normalize image_id across all annotations
+            if "image_url" in annotation:
+                # annotation["image_url"] corresponds to image_metadata["url"]
+                assert (
+                    image_metadata["url"] == annotation["image_url"]
+                ), f"Annotations doesn't match with image metadataset. Got image_metadata['url']: {image_metadata['url']} and annotations['image_url']: {annotation['image_url']}"
+                del annotation["image_url"]
+            elif "url" in annotation:
+                # annotation["url"] corresponds to image_metadata["url"]
+                assert (
+                    image_metadata["url"] == annotation["url"]
+                ), f"Annotations doesn't match with image metadataset. Got image_metadata['url']: {image_metadata['url']} and annotations['url']: {annotation['url']}"
+
+            # in-place operation to normalize annotations
+            annotation_normalizer_(annotation)
+
+            # optionally add image to the annotation
+            if image_folder_local_paths is not None:
+                filepath = _get_local_image_path(image_metadata["url"], image_folder_local_paths)
+                image_dict = {"image": filepath}
+            else:
+                image_dict = {}
+
+            yield idx, {**image_dict, **image_metadata, **annotation}