# Vision Datasets
## Introduction
This repo
- defines unified contract for dataset for purposes such as training, visualization, and exploration, via `DatasetManifest`, `ImageDataManifest`, etc.
- provides many commonly used dataset operation, such as sample dataset by categories, sample few-shot sub-dataset, sample dataset by ratios, train-test split, merge dataset, etc. (See [Here](#oom))
- provides API for organizing and accessing datasets, via `DatasetHub`
Currently, seven `basic` types of data are supported:
- `image_classification_multiclass`: each image can is only with one label.
- `image_classification_multilabel`: each image can is with one or multiple labels (e.g., 'cat', 'animal', 'pet').
- `image_object_detection`: each image is labeled with bounding boxes surrounding the objects of interest.
- `image_text_matching`: each image is associated with a collection of texts describing the image, and whether each text description matches the image or not.
- `image_matting`: each image has a pixel-wise annotation, where each pixel is labeled as 'foreground' or 'background'.
- `image_regression`: each image is labeled with a real-valued numeric regression target.
- `image_caption`: each image is labeled with a few texts describing the images.
- `text_2_image_retrieval`: each image is labeled with a number of text queries describing the image. Optionally, an image is associated with one label.
- `visual_question_answering`: each image is labeled with a number of question-answer pairs
- `visual_object_grounding`: each image is labeled with a number of question-answer-bboxes triplets.
`multitask` type is a composition type, where one set of images has multiple sets of annotations available for different tasks, where each task can be of any basic type.
**Note that `image_caption` and `text_2_image_retrieval` might be merged into `image_text_matching` in future.**
## Dataset Contracts
- `DatasetManifest` wraps the information about a dataset including labelmap, images (width, height, path to image), and annotations. `ImageDataManifest` encapsulates information about each image.
- `ImageDataManifest` encapsulates image-specific information, such as image id, path, labels, and width/height. One thing to note here is that the image path can be
1. a local path (absolute `c:\images\1.jpg` or relative `images\1.jpg`)
2. a local path in a **non-compressed** zip file (absolute `c:\images.zip@1.jpg` or relative `images.zip@1.jpg`) or
3. an url
- `ImageLabelManifest`: encapsulates one single image-level annotation
- `CategoryManifest`: encapsulates the information about a category, such as its name and super category, if applicable
- `VisionDataset` is an iterable dataset class that consumes the information from `DatasetManifest`.
`VisionDataset` is able to load the data from all three kinds of paths. Both 1. and 2. are good for training, as they access data from local disk while the 3rd one is good for data exploration, if you have the data in azure storage.
For `multitask` dataset, the labels stored in the `ImageDataManifest` is a `dict` mapping from task name to that task's labels. The labelmap stored in `DatasetManifest` is also a `dict` mapping from task name to that task's labels.
### Creating DatasetManifest
In addition to loading a serialized `DatasetManifest` for instantiation, this repo currently supports two formats of data that can instantiates `DatasetManifest`,
using `DatasetManifest.create_dataset_manifest(dataset_info, usage, container_sas_or_root_dir)`: `COCO` and `IRIS` (legacy).
`DatasetInfo` as the first arg in the arg list wraps the metainfo about the dataset like the name of the dataset, locations of the images, annotation files, etc. See examples in the sections below
for different data formats.
Once a `DatasetManifest` is created, you can create a `VisionDataset` for accessing the data in the dataset, especially the image data, for training, visualization, etc:
```{python}
dataset = VisionDataset(dataset_info, dataset_manifest, coordinates='relative')
```
#### Coco format
Here is an example with explanation of what a `DatasetInfo` looks like for coco format, when it is serialized into json:
```json
{
"name": "sampled-ms-coco",
"version": 1,
"description": "A sampled ms-coco dataset.",
"type": "object_detection",
"format": "coco", // indicating the annotation data are stored in coco format
"root_folder": "detection/coco2017_20200401", // a root folder for all files listed
"train": {
"index_path": "train.json", // coco json file for training, see next section for example
"files_for_local_usage": [ // associated files including data such as images
"images/train_images.zip"
]
},
"val": {
"index_path": "val.json",
"files_for_local_usage": [
"images/val_images.zip"
]
},
"test": {
"index_path": "test.json",
"files_for_local_usage": [
"images/test_images.zip"
]
}
}
```
Coco annotation format details w.r.t. `image_classification_multiclass/label`, `image_object_detection`, `image_caption`, `image_text_match` and `multitask` can be found in `COCO_DATA_FORMAT.md`.
Index file can be put into a zip file as well (e.g., `annotations.zip@train.json`), no need to add the this zip to "files_for_local_usage" explicitly.
#### Iris format
Iris format is a legacy format which can be found in `IRIS_DATA_FORMAT.md`. Only `multiclass/label_classification`, `object_detection` and `multitask` are supported.
## Dataset management and access
Check [DATA_PREPARATION.md](DATA_PREPARATION.md) for complete guide on how to prepare datasets in steps.
Once you have multiple datasets, it is more convenient to have all the `DatasetInfo` in one place and instantiate `DatasetManifest` or even `VisionDataset` by just using the dataset name, usage (
train, val ,test) and version.
This repo offers the class `DatasetHub` for this purpose. Once instantiated with a json including the `DatasetInfo` for all datasets, you can retrieve a `VisionDataset` by
```python
import pathlib
from vision_datasets.common import Usages, DatasetHub
dataset_infos_json_path = 'datasets.json'
dataset_hub = DatasetHub(pathlib.Path(dataset_infos_json_path).read_text(), blob_container_sas, local_dir)
stanford_cars = dataset_hub.create_manifest_dataset('stanford-cars', version=1, usage=Usages.TRAIN)
# note that you can pass multiple datasets.json to DatasetHub, it can combine them all
# example: DatasetHub([ds_json1, ds_json2, ...])
# note that you can specify multiple usages in create_manifest_dataset call
# example dataset_hub.create_manifest_dataset('stanford-cars', version=1, usage=[Usages.TRAIN, Usages.VAL])
for img, targets, sample_idx_str in stanford_cars:
img.show()
img.close()
print(targets)
```
Note that this hub class works with data saved in both Azure Blob container and on local disk.
If `local_dir`:
1. is provided, the hub will look for the resources locally and **download the data** (files included in "
files_for_local_usage", the index files, metadata (if iris format), labelmap (if iris format))
from `blob_container_sas` if not present locally
2. is NOT provided (i.e. `None`), the hub will create a manifest dataset that directly consumes data from the blob
indicated by `blob_container_sas`. Note that this does not work, if data are stored in zipped files. You will have to
unzip your data in the azure blob. (Index files requires no update, if image paths are for zip files: `a.zip@1.jpg`).
This kind of azure-based dataset is good for large dataset exploration, but can be slow for training.
When data exists on local disk, `blob_container_sas` can be `None`.
## Operations on manifests {#oom}
There are supported operations on manifests for different data types, such as split, merge, sample, etc. You can run
`vision_list_supported_operations -d {DATA_TYPE}`
to see the supported operations for a specific data type. You can use the factory classes in `vision_datasets.common.factory` to create operations for certain data type.
```python
from vision_datasets.common import DatasetTypes, SplitFactory, SplitConfig
data_manifest = ....
splitter = SplitFactory.create(DatasetTypes.IMAGE_CLASSIFICATION_MULTICLASS, SplitConfig(ratio=0.3))
manifest_1, manifest_2 = splitter.run(data_manifest)
```
### Training with PyTorch
Training with PyTorch is easy. After instantiating a `VisionDataset`, simply passing it in `vision_datasets.common.dataset.TorchDataset` together with the `transform`, then you are good to go with the PyTorch DataLoader for training.
## Helpful commands
There are a few commands that come with this repo once installed, such as datset check and download, detection conversion to classification dataset, and so on, check [`UTIL_COMMANDS.md`](./UTIL_COMMANDS.md) for details.
Raw data
{
"_id": null,
"home_page": "https://github.com/microsoft/vision-datasets",
"name": "vision-datasets",
"maintainer": "",
"docs_url": null,
"requires_python": ">=3.8",
"maintainer_email": "",
"keywords": "vision datasets classification detection",
"author": "Ping Jin, Shohei Ono",
"author_email": "",
"download_url": "https://files.pythonhosted.org/packages/e8/31/b3945be873ac19c360fdaa2f649372f24befda8b3a305940db2afa4cfd1a/vision_datasets-1.0.12.tar.gz",
"platform": null,
"description": "# Vision Datasets\n\n## Introduction\n\nThis repo\n\n- defines unified contract for dataset for purposes such as training, visualization, and exploration, via `DatasetManifest`, `ImageDataManifest`, etc.\n- provides many commonly used dataset operation, such as sample dataset by categories, sample few-shot sub-dataset, sample dataset by ratios, train-test split, merge dataset, etc. (See [Here](#oom))\n- provides API for organizing and accessing datasets, via `DatasetHub`\n\nCurrently, seven `basic` types of data are supported:\n\n- `image_classification_multiclass`: each image can is only with one label.\n- `image_classification_multilabel`: each image can is with one or multiple labels (e.g., 'cat', 'animal', 'pet').\n- `image_object_detection`: each image is labeled with bounding boxes surrounding the objects of interest.\n- `image_text_matching`: each image is associated with a collection of texts describing the image, and whether each text description matches the image or not.\n- `image_matting`: each image has a pixel-wise annotation, where each pixel is labeled as 'foreground' or 'background'.\n- `image_regression`: each image is labeled with a real-valued numeric regression target.\n- `image_caption`: each image is labeled with a few texts describing the images.\n- `text_2_image_retrieval`: each image is labeled with a number of text queries describing the image. Optionally, an image is associated with one label.\n- `visual_question_answering`: each image is labeled with a number of question-answer pairs\n- `visual_object_grounding`: each image is labeled with a number of question-answer-bboxes triplets.\n\n`multitask` type is a composition type, where one set of images has multiple sets of annotations available for different tasks, where each task can be of any basic type.\n\n**Note that `image_caption` and `text_2_image_retrieval` might be merged into `image_text_matching` in future.**\n\n## Dataset Contracts\n\n- `DatasetManifest` wraps the information about a dataset including labelmap, images (width, height, path to image), and annotations. `ImageDataManifest` encapsulates information about each image.\n- `ImageDataManifest` encapsulates image-specific information, such as image id, path, labels, and width/height. One thing to note here is that the image path can be\n 1. a local path (absolute `c:\\images\\1.jpg` or relative `images\\1.jpg`)\n 2. a local path in a **non-compressed** zip file (absolute `c:\\images.zip@1.jpg` or relative `images.zip@1.jpg`) or\n 3. an url\n- `ImageLabelManifest`: encapsulates one single image-level annotation\n- `CategoryManifest`: encapsulates the information about a category, such as its name and super category, if applicable\n- `VisionDataset` is an iterable dataset class that consumes the information from `DatasetManifest`.\n\n`VisionDataset` is able to load the data from all three kinds of paths. Both 1. and 2. are good for training, as they access data from local disk while the 3rd one is good for data exploration, if you have the data in azure storage.\n\nFor `multitask` dataset, the labels stored in the `ImageDataManifest` is a `dict` mapping from task name to that task's labels. The labelmap stored in `DatasetManifest` is also a `dict` mapping from task name to that task's labels.\n\n### Creating DatasetManifest\n\nIn addition to loading a serialized `DatasetManifest` for instantiation, this repo currently supports two formats of data that can instantiates `DatasetManifest`,\nusing `DatasetManifest.create_dataset_manifest(dataset_info, usage, container_sas_or_root_dir)`: `COCO` and `IRIS` (legacy).\n\n`DatasetInfo` as the first arg in the arg list wraps the metainfo about the dataset like the name of the dataset, locations of the images, annotation files, etc. See examples in the sections below\nfor different data formats.\n\nOnce a `DatasetManifest` is created, you can create a `VisionDataset` for accessing the data in the dataset, especially the image data, for training, visualization, etc:\n\n```{python}\ndataset = VisionDataset(dataset_info, dataset_manifest, coordinates='relative')\n```\n\n#### Coco format\n\nHere is an example with explanation of what a `DatasetInfo` looks like for coco format, when it is serialized into json:\n\n```json\n {\n \"name\": \"sampled-ms-coco\",\n \"version\": 1,\n \"description\": \"A sampled ms-coco dataset.\",\n \"type\": \"object_detection\",\n \"format\": \"coco\", // indicating the annotation data are stored in coco format\n \"root_folder\": \"detection/coco2017_20200401\", // a root folder for all files listed\n \"train\": {\n \"index_path\": \"train.json\", // coco json file for training, see next section for example\n \"files_for_local_usage\": [ // associated files including data such as images\n \"images/train_images.zip\"\n ]\n },\n \"val\": {\n \"index_path\": \"val.json\",\n \"files_for_local_usage\": [\n \"images/val_images.zip\"\n ]\n },\n \"test\": {\n \"index_path\": \"test.json\",\n \"files_for_local_usage\": [\n \"images/test_images.zip\"\n ]\n }\n }\n```\n\nCoco annotation format details w.r.t. `image_classification_multiclass/label`, `image_object_detection`, `image_caption`, `image_text_match` and `multitask` can be found in `COCO_DATA_FORMAT.md`.\n\nIndex file can be put into a zip file as well (e.g., `annotations.zip@train.json`), no need to add the this zip to \"files_for_local_usage\" explicitly.\n\n#### Iris format\n\nIris format is a legacy format which can be found in `IRIS_DATA_FORMAT.md`. Only `multiclass/label_classification`, `object_detection` and `multitask` are supported.\n\n## Dataset management and access\n\nCheck [DATA_PREPARATION.md](DATA_PREPARATION.md) for complete guide on how to prepare datasets in steps.\n\nOnce you have multiple datasets, it is more convenient to have all the `DatasetInfo` in one place and instantiate `DatasetManifest` or even `VisionDataset` by just using the dataset name, usage (\ntrain, val ,test) and version.\n\nThis repo offers the class `DatasetHub` for this purpose. Once instantiated with a json including the `DatasetInfo` for all datasets, you can retrieve a `VisionDataset` by\n\n```python\nimport pathlib\nfrom vision_datasets.common import Usages, DatasetHub\n\ndataset_infos_json_path = 'datasets.json'\ndataset_hub = DatasetHub(pathlib.Path(dataset_infos_json_path).read_text(), blob_container_sas, local_dir)\nstanford_cars = dataset_hub.create_manifest_dataset('stanford-cars', version=1, usage=Usages.TRAIN)\n\n# note that you can pass multiple datasets.json to DatasetHub, it can combine them all\n# example: DatasetHub([ds_json1, ds_json2, ...])\n# note that you can specify multiple usages in create_manifest_dataset call\n# example dataset_hub.create_manifest_dataset('stanford-cars', version=1, usage=[Usages.TRAIN, Usages.VAL])\n\nfor img, targets, sample_idx_str in stanford_cars:\n img.show()\n img.close()\n print(targets)\n```\n\nNote that this hub class works with data saved in both Azure Blob container and on local disk.\n\nIf `local_dir`:\n\n1. is provided, the hub will look for the resources locally and **download the data** (files included in \"\n files_for_local_usage\", the index files, metadata (if iris format), labelmap (if iris format))\n from `blob_container_sas` if not present locally\n2. is NOT provided (i.e. `None`), the hub will create a manifest dataset that directly consumes data from the blob\n indicated by `blob_container_sas`. Note that this does not work, if data are stored in zipped files. You will have to\n unzip your data in the azure blob. (Index files requires no update, if image paths are for zip files: `a.zip@1.jpg`).\n This kind of azure-based dataset is good for large dataset exploration, but can be slow for training.\n\nWhen data exists on local disk, `blob_container_sas` can be `None`.\n\n## Operations on manifests {#oom}\n\nThere are supported operations on manifests for different data types, such as split, merge, sample, etc. You can run\n\n`vision_list_supported_operations -d {DATA_TYPE}`\n\nto see the supported operations for a specific data type. You can use the factory classes in `vision_datasets.common.factory` to create operations for certain data type.\n\n```python\nfrom vision_datasets.common import DatasetTypes, SplitFactory, SplitConfig\n\n\ndata_manifest = ....\nsplitter = SplitFactory.create(DatasetTypes.IMAGE_CLASSIFICATION_MULTICLASS, SplitConfig(ratio=0.3))\nmanifest_1, manifest_2 = splitter.run(data_manifest)\n```\n\n### Training with PyTorch\n\nTraining with PyTorch is easy. After instantiating a `VisionDataset`, simply passing it in `vision_datasets.common.dataset.TorchDataset` together with the `transform`, then you are good to go with the PyTorch DataLoader for training.\n\n\n## Helpful commands\n\nThere are a few commands that come with this repo once installed, such as datset check and download, detection conversion to classification dataset, and so on, check [`UTIL_COMMANDS.md`](./UTIL_COMMANDS.md) for details.\n",
"bugtrack_url": null,
"license": "MIT",
"summary": "A utility repo for vision dataset access and management.",
"version": "1.0.12",
"project_urls": {
"Homepage": "https://github.com/microsoft/vision-datasets"
},
"split_keywords": [
"vision",
"datasets",
"classification",
"detection"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "589bc8a54f568f41fca25dfa779661e4a185e2e99fdb16e42f6d9a16bf5dddfa",
"md5": "29ff9a85846b1e503aa2d5822586f68d",
"sha256": "a721699752600b926235c012eb27a1b37e62931a350b6d2eb5494823c85656bb"
},
"downloads": -1,
"filename": "vision_datasets-1.0.12-py3-none-any.whl",
"has_sig": false,
"md5_digest": "29ff9a85846b1e503aa2d5822586f68d",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.8",
"size": 124793,
"upload_time": "2024-02-02T22:00:58",
"upload_time_iso_8601": "2024-02-02T22:00:58.123949Z",
"url": "https://files.pythonhosted.org/packages/58/9b/c8a54f568f41fca25dfa779661e4a185e2e99fdb16e42f6d9a16bf5dddfa/vision_datasets-1.0.12-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "e831b3945be873ac19c360fdaa2f649372f24befda8b3a305940db2afa4cfd1a",
"md5": "6b7963834dc598bd51a8eddf21148732",
"sha256": "f4ce36badcfe0f53e35404e77cfdc6405cf6f88360c6a704105547992b51ee87"
},
"downloads": -1,
"filename": "vision_datasets-1.0.12.tar.gz",
"has_sig": false,
"md5_digest": "6b7963834dc598bd51a8eddf21148732",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.8",
"size": 73729,
"upload_time": "2024-02-02T22:01:00",
"upload_time_iso_8601": "2024-02-02T22:01:00.174458Z",
"url": "https://files.pythonhosted.org/packages/e8/31/b3945be873ac19c360fdaa2f649372f24befda8b3a305940db2afa4cfd1a/vision_datasets-1.0.12.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-02-02 22:01:00",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "microsoft",
"github_project": "vision-datasets",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"tox": true,
"lcname": "vision-datasets"
}
JSON
