Skip to main content

The official Python SDK for FastLabel API, the Data Platform for AI

Project description

FastLabel Python SDK

If you are using FastLabel prototype, please install version 0.2.2.

Table of Contents

Installation

pip install --upgrade fastlabel

Python version 3.7 or greater is required

Usage

Configure API Key in environment variable.

export FASTLABEL_ACCESS_TOKEN="YOUR_ACCESS_TOKEN"

Initialize fastlabel client.

import fastlabel
client = fastlabel.Client()

Limitation

API is allowed to call 10000 times per 10 minutes. If you create/delete a large size of tasks, please wait a second for every requests.

Task

Image

Supported following project types:

  • Image - Bounding Box
  • Image - Polygon
  • Image - Keypoint
  • Image - Line
  • Image - Segmentation
  • Image - Pose Estimation
  • Image - All

Create Task

Create a new task.

task_id = client.create_image_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.jpg",
    file_path="./sample.jpg"
)

Create a new task with pre-defined annotations. (Class should be configured on your project in advance)

task_id = client.create_image_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.jpg",
    file_path="./sample.jpg",
    annotations=[{
        "type": "bbox",
        "value": "annotation-value",
        "attributes": [
            {
                "key": "attribute-key",
                "value": "attribute-value"
            }
        ],
        "points": [
            100,  # top-left x
            100,  # top-left y
            200,  # bottom-right x
            200   # bottom-right y
        ]
    }]
)

Check examples/create_image_task.py.

Limitation
  • You can upload up to a size of 20 MB.

Find Task

Find a single task.

task = client.find_image_task(task_id="YOUR_TASK_ID")

Find a single task by name.

tasks = client.find_image_task_by_name(project="YOUR_PROJECT_SLUG", task_name="YOUR_TASK_NAME")

Get Tasks

Get tasks. (Up to 1000 tasks)

tasks = client.get_image_tasks(project="YOUR_PROJECT_SLUG")
  • Filter and Get tasks. (Up to 1000 tasks)
tasks = client.get_image_tasks(
    project="YOUR_PROJECT_SLUG",
    status="approved", # status can be 'pending', 'registered', 'completed', 'skipped', 'reviewed' 'sent_back', 'approved', 'declined'
    tags=["tag1", "tag2"] # up to 10 tags
)

Get a large size of tasks. (Over 1000 tasks)

import time

# Iterate pages until new tasks are empty.
all_tasks = []
offset = None
while True:
    time.sleep(1)

    tasks = client.get_image_tasks(project="YOUR_PROJECT_SLUG", offset=offset)
    all_tasks.extend(tasks)

    if len(tasks) > 0:
        offset = len(all_tasks)  # Set the offset
    else:
        break

Please wait a second before sending another requests!

Update Tasks

Update a single task.

task_id = client.update_image_task(
    task_id="YOUR_TASK_ID",
    status="approved",
    assignee="USER_SLUG",
    tags=["tag1", "tag2"],
    annotations=[
        {
            "type": "bbox",
            "value": "cat"
            "attributes": [
                { "key": "kind", "value": "Scottish field" }
            ],
            "points": [
                100,  # top-left x
                100,  # top-left y
                200,  # bottom-right x
                200   # bottom-right y
            ]
        }
    ],
)

Response

Example of a single image task object

{
    "id": "YOUR_TASK_ID",
    "name": "cat.jpg",
    "width": 100,   # image width
    "height": 100,  # image height
    "url": "YOUR_TASK_URL",
    "status": "registered",
    "externalStatus": "registered",
    "tags": [],
    "assignee": "ASSIGNEE_NAME",
    "reviewer": "REVIEWER_NAME",
    "externalAssignee": "EXTERNAL_ASSIGNEE_NAME",
    "externalReviewer": "EXTERNAL_REVIEWER_NAME",
    "annotations": [
        {
            "attributes": [
                { "key": "kind", "name": "Kind", "type": "text", "value": "Scottish field" }
            ],
            "color": "#b36d18",
            "points": [
                100,  # top-left x
                100,  # top-left y
                200,  # bottom-right x
                200   # bottom-right y
            ],
            "rotation": 0,
            "title": "Cat",
            "type": "bbox",
            "value": "cat"
        }
    ],
    "createdAt": "2021-02-22T11:25:27.158Z",
    "updatedAt": "2021-02-22T11:25:27.158Z"
}

Example when the project type is Image - Pose Estimation

{
    "id": "YOUR_TASK_ID",
    "name": "person.jpg",
    "width": 255,   # image width
    "height": 255,  # image height
    "url": "YOUR_TASK_URL",
    "status": "registered",
    "externalStatus": "registered",
    "tags": [],
    "assignee": "ASSIGNEE_NAME",
    "reviewer": "REVIEWER_NAME",
    "externalAssignee": "EXTERNAL_ASSIGNEE_NAME",
    "externalReviewer": "EXTERNAL_REVIEWER_NAME",
    "annotations":[
       {
          "type":"pose_estimation",
          "title":"jesture",
          "value":"jesture",
          "color":"#10c414",
          "attributes": [],
          "keypoints":[
             {
                "name":"頭",
                "key":"head",
                "value":[
                   102.59, # x
                   23.04,  # y
                   1       # 0:invisible, 1:visible
                ],
                "edges":[
                   "right_shoulder",
                   "left_shoulder"
                ]
             },
             {
                "name":"右肩",
                "key":"right_shoulder",
                "value":[
                   186.69,
                   114.11,
                   1
                ],
                "edges":[
                   "head"
                ]
             },
             {
                "name":"左肩",
                "key":"left_shoulder",
                "value":[
                   37.23,
                   109.29,
                   1
                ],
                "edges":[
                   "head"
                ]
             }
          ]
       }
    ],
    "createdAt": "2021-02-22T11:25:27.158Z",
    "updatedAt": "2021-02-22T11:25:27.158Z"
}

Export Image With Annotations

Get tasks and export images with annotations. Only support the following image extension.

  • jpeg
  • jpg
  • png
  • tif
  • tiff
  • bmp
tasks = client.get_image_tasks(project="YOUR_PROJECT_SLUG")
client.export_image_with_annotations(
    tasks=tasks, image_dir="IMAGE_DIR", output_dir="OUTPUT_DIR"
)

Integrate Task

This function is alpha version. It is subject to major changes in the future.

Integration is possible only when tasks are registered from the objects divided by the dataset. Only bbox and polygon annotation types are supported.

In the case of a task divided under the following conditions.

  • Dataset slug: image
  • Object name: cat.jpg
  • Split count: 3×3

Objects are registered in the data set in the following form.

  • image/cat/1.jpg
  • image/cat/2.jpg
  • image/cat/3.jpg
  • (omit)
  • image/cat/9.jpg

The annotations at the edges of the image are combined. However, annotations with a maximum length of 300px may not work.

In this case, SPLIT_IMAGE_TASK_NAME_PREFIX specifies image/cat.

task = client.find_integrated_image_task_by_prefix(
    project="YOUR_PROJECT_SLUG", 
    prefix="SPLIT_IMAGE_TASK_NAME_PREFIX",
)
Response

Example of a integrated image task object

{
    'name': 'image/cat.jpg',
    "annotations": [
        {
            "attributes": [],
            "color": "#b36d18",
            "confidenceScore"; -1,
            "keypoints": [],
            "points": [200,200,300,400],
            "rotation": 0,
            "title": "Bird",
            "type": "polygon",
            "value": "bird"
        }
    ],
}

Image Classification

Supported following project types:

  • Image - Classification

Create Task

Create a new task.

task_id = client.create_image_classification_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.jpg",
    file_path="./sample.jpg",
    attributes=[
        {
            "key": "attribute-key",
            "value": "attribute-value"
        }
    ],
)
Limitation
  • You can upload up to a size of 20 MB.

Find Task

Find a single task.

task = client.find_image_classification_task(task_id="YOUR_TASK_ID")

Find a single task by name.

tasks = client.find_image_classification_task_by_name(project="YOUR_PROJECT_SLUG", task_name="YOUR_TASK_NAME")

Get Tasks

Get tasks. (Up to 1000 tasks)

tasks = client.get_image_classification_tasks(project="YOUR_PROJECT_SLUG")

Update Tasks

Update a single task.

task_id = client.update_image_classification_task(
    task_id="YOUR_TASK_ID",
    status="approved",
    assignee="USER_SLUG",
    tags=["tag1", "tag2"],
    attributes=[
        {
            "key": "attribute-key",
            "value": "attribute-value"
        }
    ],
)

Response

Example of a single image classification task object

{
    "id": "YOUR_TASK_ID",
    "name": "cat.jpg",
    "width": 100,   # image width
    "height": 100,  # image height
    "url": "YOUR_TASK_URL",
    "status": "registered",
    "externalStatus": "registered",
    "tags": [],
    "assignee": "ASSIGNEE_NAME",
    "reviewer": "REVIEWER_NAME",
    "externalAssignee": "EXTERNAL_ASSIGNEE_NAME",
    "externalReviewer": "EXTERNAL_REVIEWER_NAME",
    "attributes": [
        {
            "key": "kind",
            "name": "Kind",
            "type": "text",
            "value": "Scottish field"
        }
    ],
    "createdAt": "2021-02-22T11:25:27.158Z",
    "updatedAt": "2021-02-22T11:25:27.158Z"
}

Multi Image

Supported following project types:

  • Multi Image - Bounding Box
  • Multi Image - Polygon
  • Multi Image - Keypoint
  • Multi Image - Line
  • Multi Image - Segmentation

Create Task

Create a new task.

task = client.create_multi_image_task(
    project="YOUR_PROJECT_SLUG",
    name="sample",
    folder_path="./sample",
    annotations=[{
        "type": "segmentation",
        "value": "annotation-value",
        "attributes": [
            {
                "key": "attribute-key",
                "value": "attribute-value"
            }
        ],
        "content": "01.jpg",
        "points": [[[
            100,
            100,
            300,
            100,
            300,
            300,
            100,
            300,
            100,
            100
        ]]] # clockwise rotation
    }]
)
Limitation
  • You can upload up to a size of 20 MB.
  • You can upload up to a total size of 512 MB.
  • You can upload up to 250 files in total.

Find Task

Find a single task.

task = client.find_multi_image_task(task_id="YOUR_TASK_ID")

Find a single task by name.

tasks = client.find_multi_image_task_by_name(project="YOUR_PROJECT_SLUG", task_name="YOUR_TASK_NAME")

Get Tasks

Get tasks.

tasks = client.get_multi_image_tasks(project="YOUR_PROJECT_SLUG")

Update Task

Update a single task.

task_id = client.update_multi_image_task(
    task_id="YOUR_TASK_ID",
    status="approved",
    assignee="USER_SLUG",
    tags=["tag1", "tag2"],
    annotations=[
        {
            "type": "bbox",
            "value": "cat",
            "content": "cat1.jpg",
            "attributes": [
                { "key": "key", "value": "value1" }
            ],
            "points": [990, 560, 980, 550]
        }
    ]
)

Response

Example of a single task object

{
    "id": "YOUR_TASK_ID",
    "name": "cat.jpg",
    "contents": [
        {
            "name": "content-name",
            "url": "content-url",
            "width": 100,
            "height": 100,
        }
    ],
    "status": "registered",
    "externalStatus": "registered",
    "tags": [],
    "assignee": "ASSIGNEE_NAME",
    "reviewer": "REVIEWER_NAME",
    "externalAssignee": "EXTERNAL_ASSIGNEE_NAME",
    "externalReviewer": "EXTERNAL_REVIEWER_NAME",
    "annotations": [
        {
            "content": "content-name"
            "attributes": [],
            "color": "#b36d18",
            "points": [[[
                100,
                100,
                300,
                100,
                300,
                300,
                100,
                300,
                100,
                100
            ]]]
            "title": "Cat",
            "type": "bbox",
            "value": "cat"
        }
    ],
    "createdAt": "2021-02-22T11:25:27.158Z",
    "updatedAt": "2021-02-22T11:25:27.158Z"
}

Video

Supported following project types:

  • Video - Bounding Box
  • Video - Keypoint
  • Video - Line

Create Task

Create a new task.

task_id = client.create_video_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.mp4",
    file_path="./sample.mp4"
)

Create a new task with pre-defined annotations. (Class should be configured on your project in advance)

task_id = client.create_video_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.mp4",
    file_path="./sample.mp4",
    annotations=[{
        "type": "bbox",
        "value": "person",
        "points": {
            "1": {  # number of frame
                "value": [
                    100,  # top-left x
                    100,  # top-left y
                    200,  # bottom-right x
                    200   # bottom-right y
                ],
                # Make sure to set `autogenerated` False for the first and last frame. "1" and "3" frames in this case.
                # Otherwise, annotation is auto-completed for rest of frames when you edit.
                "autogenerated": False
            },
            "2": {
                "value": [
                    110,
                    110,
                    220,
                    220
                ],
                "autogenerated": True
            },
            "3": {
                "value": [
                    120,
                    120,
                    240,
                    240
                ],
                "autogenerated": False
            }
        }
    }]
)
Limitation
  • You can upload up to a size of 250 MB.

Find Task

Find a single task.

task = client.find_video_task(task_id="YOUR_TASK_ID")

Find a single task by name.

tasks = client.find_video_task_by_name(project="YOUR_PROJECT_SLUG", task_name="YOUR_TASK_NAME")

Get Tasks

Get tasks. (Up to 10 tasks)

tasks = client.get_video_tasks(project="YOUR_PROJECT_SLUG")

Update Task

Update a single task.

task_id = client.update_video_task(
    task_id="YOUR_TASK_ID",
    status="approved",
    assignee="USER_SLUG",
    tags=["tag1", "tag2"],
    annotations=[{
        "type": "bbox",
        "value": "bird",
        "points": {
            "1": { 
                "value": [
                    100,
                    100,
                    200,
                    200
                ],
                "autogenerated": False
            },
            "2": {
                "value": [
                    110,
                    110,
                    220,
                    220
                ],
                "autogenerated": True
            },
            "3": {
                "value": [
                    120,
                    120,
                    240,
                    240
                ],
                "autogenerated": False
            }
        }
    }]
)

Integrate Video

This function is alpha version. It is subject to major changes in the future.

Integration is possible only when tasks are registered from the objects divided by the dataset.

In the case of a task divided under the following conditions.

  • Dataset slug: video
  • Object name: cat.mp4
  • Split count: 3

Objects are registered in the data set in the following form.

  • video/cat/1.mp4
  • video/cat/2.mp4
  • video/cat/3.mp4

In this case, SPLIT_VIDEO_TASK_NAME_PREFIX specifies video/cat.

task = client.find_integrated_video_task_by_prefix(
    project="YOUR_PROJECT_SLUG", 
    prefix="SPLIT_VIDEO_TASK_NAME_PREFIX",
)

Response

Example of a single vide task object

{
    "id": "YOUR_TASK_ID",
    "name": "cat.jpg",
    "width": 100,   # image width
    "height": 100,  # image height
    "fps": 30.0,    # frame per seconds
    "frameCount": 480,  # total frame count of video
    "duration": 16.0,   # total duration of video
    "url": "YOUR_TASK_URL",
    "status": "registered",
    "externalStatus": "registered",
    "tags": [],
    "assignee": "ASSIGNEE_NAME",
    "reviewer": "REVIEWER_NAME",
    "externalAssignee": "EXTERNAL_ASSIGNEE_NAME",
    "externalReviewer": "EXTERNAL_REVIEWER_NAME",
    "annotations": [
        {
            "attributes": [],
            "color": "#b36d18",
            "points": {
                "1": {  # number of frame
                    "value": [
                        100,  # top-left x
                        100,  # top-left y
                        200,  # bottom-right x
                        200   # bottom-right y
                    ],
                    "autogenerated": False  # False when annotated manually. True when auto-generated by system.
                },
                "2": {
                    "value": [
                        110,
                        110,
                        220,
                        220
                    ],
                    "autogenerated": True
                },
                "3": {
                    "value": [
                        120,
                        120,
                        240,
                        240
                    ],
                    "autogenerated": False
                }
            },
            "title": "Cat",
            "type": "bbox",
            "value": "cat"
        }
    ],
    "createdAt": "2021-02-22T11:25:27.158Z",
    "updatedAt": "2021-02-22T11:25:27.158Z"
}

Video Classification

Supported following project types:

  • Video - Classification (Single)

Create Task

Create a new task.

task_id = client.create_video_classification_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.mp4",
    file_path="./sample.mp4",
    attributes=[
        {
            "key": "attribute-key",
            "value": "attribute-value"
        }
    ],
)
Limitation
  • You can upload up to a size of 250 MB.

Find Task

Find a single task.

task = client.find_video_classification_task(task_id="YOUR_TASK_ID")

Find a single task by name.

tasks = client.find_video_classification_task_by_name(project="YOUR_PROJECT_SLUG", task_name="YOUR_TASK_NAME")

Get Tasks

Get tasks. (Up to 1000 tasks)

tasks = client.get_video_classification_tasks(project="YOUR_PROJECT_SLUG")

Update Tasks

Update a single task.

task_id = client.update_video_classification_task(
    task_id="YOUR_TASK_ID",
    status="approved",
    assignee="USER_SLUG",
    tags=["tag1", "tag2"],
    attributes=[
        {
            "key": "attribute-key",
            "value": "attribute-value"
        }
    ],
)

Text

Supported following project types:

  • Text - NER

Create Task

Create a new task.

task_id = client.create_text_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.txt",
    file_path="./sample.txt"
)

Create a new task with pre-defined annotations. (Class should be configured on your project in advance)

task_id = client.create_text_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.txt",
    file_path="./sample.txt",
    annotations=[{
        "type": "ner",
        "value": "person",
        "start": 0,
        "end": 10,
        "text": "1234567890"
    }]
)
Limitation
  • You can upload up to a size of 2 MB.

Find Task

Find a single task.

task = client.find_text_task(task_id="YOUR_TASK_ID")

Find a single task by name.

tasks = client.find_text_task_by_name(project="YOUR_PROJECT_SLUG", task_name="YOUR_TASK_NAME")

Get Tasks

Get tasks. (Up to 10 tasks)

tasks = client.get_text_tasks(project="YOUR_PROJECT_SLUG")

Update Task

Update a single task.

task_id = client.update_text_task(
    task_id="YOUR_TASK_ID",
    status="approved",
    assignee="USER_SLUG",
    tags=["tag1", "tag2"],
    annotations=[{
        "type": "bbox",
        "value": "bird",
        "start": 0,
        "end": 10,
        "text": "0123456789"
    }]
)

Response

Example of a single text task object

{
    "id": "YOUR_TASK_ID",
    "name": "cat.txt",
    "url": "YOUR_TASK_URL",
    "status": "registered",
    "externalStatus": "registered",
    "tags": [],
    "assignee": "ASSIGNEE_NAME",
    "reviewer": "REVIEWER_NAME",
    "externalAssignee": "EXTERNAL_ASSIGNEE_NAME",
    "externalReviewer": "EXTERNAL_REVIEWER_NAME",
    "annotations": [
        {
            "attributes": [],
            "color": "#b36d18",
            "text": "0123456789",
            "start": 0,
            "end": 10,
            "title": "Cat",
            "type": "ner",
            "value": "cat"
        }
    ],
    "createdAt": "2021-02-22T11:25:27.158Z",
    "updatedAt": "2021-02-22T11:25:27.158Z"
}

Text Classification

Supported following project types:

  • Text - Classification (Single)

Create Task

Create a new task.

task_id = client.create_text_classification_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.txt",
    file_path="./sample.txt",
    attributes=[
        {
            "key": "attribute-key",
            "value": "attribute-value"
        }
    ],
)
Limitation
  • You can upload up to a size of 2 MB.

Find Task

Find a single task.

task = client.find_text_classification_task(task_id="YOUR_TASK_ID")

Find a single task by name.

tasks = client.find_text_classification_task_by_name(project="YOUR_PROJECT_SLUG", task_name="YOUR_TASK_NAME")

Get Tasks

Get tasks. (Up to 1000 tasks)

tasks = client.get_text_classification_tasks(project="YOUR_PROJECT_SLUG")

Update Tasks

Update a single task.

task_id = client.update_text_classification_task(
    task_id="YOUR_TASK_ID",
    status="approved",
    assignee="USER_SLUG",
    tags=["tag1", "tag2"],
    attributes=[
        {
            "key": "attribute-key",
            "value": "attribute-value"
        }
    ],
)

Audio

Supported following project types:

  • Audio - Segmentation

Create Task

Create a new task.

task_id = client.create_audio_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.mp3",
    file_path="./sample.mp3"
)

Create a new task with pre-defined annotations. (Class should be configured on your project in advance)

task_id = client.create_audio_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.mp3",
    file_path="./sample.mp3",
    annotations=[{
        "type": "segmentation",
        "value": "person",
        "start": 0.4,
        "end": 0.5
    }]
)
Limitation
  • You can upload up to a size of 120 MB.

Find Task

Find a single task.

task = client.find_audio_task(task_id="YOUR_TASK_ID")

Find a single task by name.

tasks = client.find_audio_task_by_name(project="YOUR_PROJECT_SLUG", task_name="YOUR_TASK_NAME")

Get Tasks

Get tasks. (Up to 10 tasks)

tasks = client.get_audio_tasks(project="YOUR_PROJECT_SLUG")

Update Task

Update a single task.

task_id = client.update_audio_task(
    task_id="YOUR_TASK_ID",
    status="approved",
    assignee="USER_SLUG",
    tags=["tag1", "tag2"],
    annotations=[{
        "type": "segmentation",
        "value": "bird",
        "start": 0.4,
        "end": 0.5
    }]
)

Response

Example of a single audio task object

{
    "id": "YOUR_TASK_ID",
    "name": "cat.mp3",
    "url": "YOUR_TASK_URL",
    "status": "registered",
    "externalStatus": "registered",
    "tags": [],
    "assignee": "ASSIGNEE_NAME",
    "reviewer": "REVIEWER_NAME",
    "externalAssignee": "EXTERNAL_ASSIGNEE_NAME",
    "externalReviewer": "EXTERNAL_REVIEWER_NAME",
    "annotations": [
        {
            "attributes": [],
            "color": "#b36d18",
            "start": 0.4,
            "end": 0.5,
            "title": "Bird",
            "type": "segmentation",
            "value": "bird"
        }
    ],
    "createdAt": "2021-02-22T11:25:27.158Z",
    "updatedAt": "2021-02-22T11:25:27.158Z"
}

Integrate Task

This function is alpha version. It is subject to major changes in the future.

Integration is possible only when tasks are registered from the objects divided by the dataset.

In the case of a task divided under the following conditions.

  • Dataset slug: audio
  • Object name: voice.mp3
  • Split count: 3

Objects are registered in the data set in the following form.

  • audio/voice/1.mp3
  • audio/voice/2.mp3
  • audio/voice/3.mp3

Annotations are combined when the end point specified in the annotation is the end time of the task and the start point of the next task is 0 seconds.

In this case, SPLIT_AUDIO_TASK_NAME_PREFIX specifies audio/voice.

task = client.find_integrated_audio_task_by_prefix(
    project="YOUR_PROJECT_SLUG", 
    prefix="SPLIT_AUDIO_TASK_NAME_PREFIX",
)
Response

Example of a integrated audio task object

{
    'name': 'audio/voice.mp3',
    "annotations": [
        {
            "attributes": [],
            "color": "#b36d18",
            "start": 0.4,
            "end": 0.5,
            "title": "Bird",
            "type": "segmentation",
            "value": "bird"
        }
    ],
}

Audio Classification

Supported following project types:

  • Audio - Classification (Single)

Create Task

Create a new task.

task_id = client.create_audio_classification_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.mp3",
    file_path="./sample.mp3",
    attributes=[
        {
            "key": "attribute-key",
            "value": "attribute-value"
        }
    ],
)
Limitation
  • You can upload up to a size of 120 MB.

Find Task

Find a single task.

task = client.find_audio_classification_task(task_id="YOUR_TASK_ID")

Find a single task by name.

tasks = client.find_audio_classification_task_by_name(project="YOUR_PROJECT_SLUG", task_name="YOUR_TASK_NAME")

Get Tasks

Get tasks. (Up to 1000 tasks)

tasks = client.get_audio_classification_tasks(project="YOUR_PROJECT_SLUG")

Update Tasks

Update a single task.

task_id = client.update_audio_classification_task(
    task_id="YOUR_TASK_ID",
    status="approved",
    assignee="USER_SLUG",
    tags=["tag1", "tag2"],
    attributes=[
        {
            "key": "attribute-key",
            "value": "attribute-value"
        }
    ],
)

Common

APIs for update and delete are same over all tasks.

Update Task

Update a single task status, tags and assignee.

task_id = client.update_task(
    task_id="YOUR_TASK_ID",
    status="approved",
    tags=["tag1", "tag2"],
    assignee="USER_SLUG"
)

Delete Task

Delete a single task.

client.delete_task(task_id="YOUR_TASK_ID")

Get Tasks Id and Name map

id_name_map = client.get_task_id_name_map(project="YOUR_PROJECT_SLUG")

Create Task from S3

Task creation from S3.

status = client.update_aws_s3_storage(
    project="YOUR_PROJECT_SLUG",
    bucket_name="S3_BUCKET_NAME",
    bucket_region="S3_REGIONS",
)
  • Run create task from AWS S3
history = client.create_task_from_aws_s3(
    project="YOUR_PROJECT_SLUG",
)
  • Get AWS S3 import status
history = client.get_aws_s3_import_status_by_project(
    project="YOUR_PROJECT_SLUG",
)

Annotation

Create Annotation

Create a new annotation.

annotation_id = client.create_annotation(
    project="YOUR_PROJECT_SLUG", type="bbox", value="cat", title="Cat")

Create a new annotation with color and attributes.

attributes = [
    {
        "type": "text",
        "name": "Kind",
        "key": "kind"
    },
    {
        "type": "select",
        "name": "Size",
        "key": "size",
        "options": [ # select, radio and checkbox type requires options
            {
                "title": "Large",
                "value": "large"
            },
            {
                "title": "Small",
                "value": "small"
            },
        ]
    },
]
annotation_id = client.create_annotation(
    project="YOUR_PROJECT_SLUG", type="bbox", value="cat", title="Cat", color="#FF0000", attributes=attributes)

Create a new classification annotation.

annotation_id = client.create_classification_annotation(
    project="YOUR_PROJECT_SLUG", attributes=attributes)

Find Annotation

Find an annotation.

annotation = client.find_annotation(annotation_id="YOUR_ANNOTATION_ID")

Find an annotation by value.

annotation = client.find_annotation_by_value(project="YOUR_PROJECT_SLUG", value="cat")

Find an annotation by value in classification project.

annotation = client.find_annotation_by_value(
    project="YOUR_PROJECT_SLUG", value="classification") # "classification" is fixed value

Get Annotations

Get annotations. (Up to 1000 annotations)

annotations = client.get_annotations(project="YOUR_PROJECT_SLUG")

Response

Example of an annotation object

{
    "id": "YOUR_ANNOTATION_ID",
    "type": "bbox",
    "value": "cat",
    "title": "Cat",
    "color": "#FF0000",
    "order": 1,
    "vertex": 0,
    "attributes": [
        {
            "id": "YOUR_ATTRIBUTE_ID",
            "key": "kind",
            "name": "Kind",
            "options": [],
            "order": 1,
            "type": "text",
            "value": ""
        },
        {
            "id": "YOUR_ATTRIBUTE_ID",
            "key": "size",
            "name": "Size",
            "options": [
                {"title": "Large", "value": "large"},
                {"title": "Small", "value": "small"}
            ],
            "order": 2,
            "type": "select",
            "value": ""
        }
    ],
    "createdAt": "2021-05-25T05:36:50.459Z",
    "updatedAt": "2021-05-25T05:36:50.459Z"
}

Example when the annotation type is Pose Estimation

{
   "id":"b12c81c3-ddec-4f98-b41b-cef7f77d26a4",
   "type":"pose_estimation",
   "title":"jesture",
   "value":"jesture",
   "color":"#10c414",
   "order":1,
   "attributes": [],
   "keypoints":[
      {
         "id":"b03ea998-a2f1-4733-b7e9-78cdf73bd38a",
         "name":"頭",
         "key":"head",
         "color":"#0033CC",
         "edges":[
            "195f5852-c516-498b-b392-24513ce3ea67",
            "06b5c968-1786-4d75-a719-951e915e5557"
         ],
         "value": []
      },
      {
         "id":"195f5852-c516-498b-b392-24513ce3ea67",
         "name":"右肩",
         "key":"right_shoulder",
         "color":"#0033CC",
         "edges":[
            "b03ea998-a2f1-4733-b7e9-78cdf73bd38a"
         ],
         "value": []
      },
      {
         "id":"06b5c968-1786-4d75-a719-951e915e5557",
         "name":"左肩",
         "key":"left_shoulder",
         "color":"#0033CC",
         "edges":[
            "b03ea998-a2f1-4733-b7e9-78cdf73bd38a"
         ],
         "value": []
      }
   ],
   "createdAt":"2021-11-21T09:59:46.714Z",
   "updatedAt":"2021-11-21T09:59:46.714Z"
}

Update Annotation

Update an annotation.

annotation_id = client.update_annotation(
    annotation_id="YOUR_ANNOTATION_ID", value="cat2", title="Cat2", color="#FF0000")

Update an annotation with attributes.

attributes = [
    {
        "id": "YOUR_ATTRIBUTE_ID",  # check by sdk get methods
        "type": "text",
        "name": "Kind2",
        "key": "kind2"
    },
    {
        "id": "YOUR_ATTRIBUTE_ID",
        "type": "select",
        "name": "Size2",
        "key": "size2",
        "options": [
            {
                "title": "Large2",
                "value": "large2"
            },
            {
                "title": "Small2",
                "value": "small2"
            },
        ]
    },
]
annotation_id = client.update_annotation(
    annotation_id="YOUR_ANNOTATION_ID", value="cat2", title="Cat2", color="#FF0000", attributes=attributes)

Update a classification annotation.

annotation_id = client.update_classification_annotation(
    project="YOUR_PROJECT_SLUG", attributes=attributes)

Delete Annotation

Delete an annotation.

client.delete_annotation(annotation_id="YOUR_ANNOTATION_ID")

Project

Create Project

Create a new project.

project_id = client.create_project(
    type="image_bbox", name="ImageNet", slug="image-net")

Find Project

Find a project.

project = client.find_project(project_id="YOUR_PROJECT_ID")

Find a project by slug.

project = client.find_project_by_slug(slug="YOUR_PROJECT_SLUG")

Get Projects

Get projects. (Up to 1000 projects)

projects = client.get_projects()

Response

Example of a project object

{
    "id": "YOUR_PROJECT_ID",
    "type": "image_bbox",
    "slug": "YOUR_PROJECT_SLUG",
    "name": "YOUR_PROJECT_NAME",
    "isPixel": False,
    "jobSize": 10,
    "status": "active",
    "createdAt": "2021-04-20T03:20:41.427Z",
    "updatedAt": "2021-04-20T03:20:41.427Z",
}

Update Project

Update a project.

project_id = client.update_project(
    project_id="YOUR_PROJECT_ID", name="NewImageNet", slug="new-image-net", job_size=20)

Delete Project

Delete a project.

client.delete_project(project_id="YOUR_PROJECT_ID")

Copy Project

Copy a project.

project_id = client.copy_project(project_id="YOUR_PROJECT_ID")

Converter

COCO

Support the following annotation types.

  • bbox
  • polygon
  • pose estimation

Get tasks and export as a COCO format file.

tasks = client.get_image_tasks(project="YOUR_PROJECT_SLUG")
client.export_coco(tasks)

Export with specifying output directory and file name.

client.export_coco(tasks=tasks, output_dir="YOUR_DIRECTROY", output_file_name="YOUR_FILE_NAME")

If you would like to export pose estimation type annotations, please pass annotations.

project_slug = "YOUR_PROJECT_SLUG"
tasks = client.get_image_tasks(project=project_slug)
annotations = client.get_annotations(project=project_slug)
client.export_coco(tasks=tasks, annotations=annotations, output_dir="YOUR_DIRECTROY", output_file_name="YOUR_FILE_NAME")

YOLO

Support the following annotation types.

  • bbox
  • polygon

Get tasks and export as YOLO format files.

tasks = client.get_image_tasks(project="YOUR_PROJECT_SLUG")
client.export_yolo(tasks, output_dir="YOUR_DIRECTROY")

Get tasks and export as YOLO format files with classes.txt You can use fixed classes.txt and arrange order of each annotaiton file's order

project_slug = "YOUR_PROJECT_SLUG"
tasks = client.get_image_tasks(project=project_slug)
annotations = client.get_annotations(project=project_slug)
classes = list(map(lambda annotation: annotation["value"], annotations))
client.export_yolo(tasks=tasks, classes=classes, output_dir="YOUR_DIRECTROY")

Pascal VOC

Support the following annotation types.

  • bbox
  • polygon

Get tasks and export as Pascal VOC format files.

tasks = client.get_image_tasks(project="YOUR_PROJECT_SLUG")
client.export_pascalvoc(tasks)

labelme

Support the following annotation types.

  • bbox
  • polygon
  • points
  • line

Get tasks and export as labelme format files.

tasks = client.get_image_tasks(project="YOUR_PROJECT_SLUG")
client.export_labelme(tasks)

Segmentation

Get tasks and export index color instance/semantic segmentation (PNG files). Only support the following annotation types.

  • bbox
  • polygon
  • segmentation (Hollowed points are not supported.)
tasks = client.get_image_tasks(project="YOUR_PROJECT_SLUG")
client.export_instance_segmentation(tasks)
tasks = client.get_image_tasks(project="YOUR_PROJECT_SLUG")
client.export_semantic_segmentation(tasks)

Converter to FastLabel format

Response

Example of a converted annotations

{
  'sample1.jpg':  [
    {
      'points': [
        100,
        100,
        200,
        200
      ],
      'type': 'bbox',
      'value': 'cat'
    }
  ],
  'sample2.jpg':  [
    {
      'points': [
        100,
        100,
        200,
        200
      ],
      'type': 'bbox',
      'value': 'cat'
    }
  ]
}

In the case of YOLO, Pascal VOC, and labelme, the key is the tree structure if the tree structure is multi-level.

dataset
├── sample1.jpg
├── sample1.txt
└── sample_dir
    ├── sample2.jpg
    └── sample2.txt
{
  'sample1.jpg':  [
    {
      'points': [
        100,
        100,
        200,
        200
      ],
      'type': 'bbox',
      'value': 'cat'
    }
  ],
  'sample_dir/sample2.jpg':  [
    {
      'points': [
        100,
        100,
        200,
        200
      ],
      'type': 'bbox',
      'value': 'cat'
    }
  ]
}

COCO

Supported bbox , polygon or pose_estimation annotation type.

Convert annotation file of COCO format as a Fastlabel format and create task.

file_path: COCO annotation json file path

annotations_map = client.convert_coco_to_fastlabel(file_path="./sample.json", annotation_type="bbox")
# annotation_type = "bbox", "polygon" or "pose_estimation 

task_id = client.create_image_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.jpg",
    file_path="./sample.jpg",
    annotations=annotations_map.get("sample.jpg")
)

Example of converting annotations to create multiple tasks.

In the case of the following tree structure.

dataset
├── annotation.json
├── sample1.jpg
└── sample2.jpg

Example source code.

import fastlabel

project = "YOUR_PROJECT_SLUG"
input_file_path = "./dataset/annotation.json"
input_dataset_path = "./dataset/"

annotations_map = client.convert_coco_to_fastlabel(file_path=input_file_path)
for image_file_path in glob.iglob(os.path.join(input_dataset_path, "**/**.jpg"), recursive=True):
    time.sleep(1)
    name = image_file_path.replace(os.path.join(*[input_dataset_path, ""]), "")
    file_path = image_file_path
    annotations = annotations_map.get(name) if annotations_map.get(name) is not None else []
    task_id = client.create_image_task(
        project=project,
        name=name,
        file_path=file_path,
        annotations=annotations
    )

YOLO

Supported bbox annotation type.

Convert annotation file of YOLO format as a Fastlabel format and create task.

classes_file_path: YOLO classes text file path
dataset_folder_path: Folder path containing YOLO Images and annotation

annotations_map = client.convert_yolo_to_fastlabel(
    classes_file_path="./classes.txt",
    dataset_folder_path="./dataset/"
)
task_id = client.create_image_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.jpg",
    file_path="./dataset/sample.jpg",
    annotations=annotations_map.get("sample.jpg")
)

Example of converting annotations to create multiple tasks.

In the case of the following tree structure.

yolo
├── classes.txt
└── dataset
    ├── sample1.jpg
    ├── sample1.txt
    ├── sample2.jpg
    └── sample2.txt

Example source code.

import fastlabel

project = "YOUR_PROJECT_SLUG"
input_file_path = "./classes.txt"
input_dataset_path = "./dataset/"
annotations_map = client.convert_yolo_to_fastlabel(
    classes_file_path=input_file_path,
    dataset_folder_path=input_dataset_path
)
for image_file_path in glob.iglob(os.path.join(input_dataset_path, "**/**.jpg"), recursive=True):
    time.sleep(1)
    name = image_file_path.replace(os.path.join(*[input_dataset_path, ""]), "")
    file_path = image_file_path
    annotations = annotations_map.get(name) if annotations_map.get(name) is not None else []
    task_id = client.create_image_task(
        project=project,
        name=name,
        file_path=file_path,
        annotations=annotations
    )

Pascal VOC

Supported bbox annotation type.

Convert annotation file of Pascal VOC format as a Fastlabel format and create task.

folder_path: Folder path including pascal VOC format annotation files

annotations_map = client.convert_pascalvoc_to_fastlabel(folder_path="./dataset/")
task_id = client.create_image_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.jpg",
    file_path="./dataset/sample.jpg",
    annotations=annotations_map.get("sample.jpg")
)

Example of converting annotations to create multiple tasks.

In the case of the following tree structure.

dataset
├── sample1.jpg
├── sample1.xml
├── sample2.jpg
└── sample2.xml

Example source code.

import fastlabel

project = "YOUR_PROJECT_SLUG"
input_dataset_path = "./dataset/"

annotations_map = client.convert_pascalvoc_to_fastlabel(folder_path=input_dataset_path)
for image_file_path in glob.iglob(os.path.join(input_dataset_path, "**/**.jpg"), recursive=True):
    time.sleep(1)
    name = image_file_path.replace(os.path.join(*[input_dataset_path, ""]), "")
    file_path = image_file_path
    annotations = annotations_map.get(name) if annotations_map.get(name) is not None else []
    task_id = client.create_image_task(
        project=project,
        name=name,
        file_path=file_path,
        annotations=annotations
    )

labelme

Support the following annotation types.

  • bbox
  • polygon
  • points
  • line

Convert annotation file of labelme format as a Fastlabel format and create task.

folder_path: Folder path including labelme format annotation files

annotations_map = client.convert_labelme_to_fastlabel(folder_path="./dataset/")
task_id = client.create_image_task(
    project="YOUR_PROJECT_SLUG",
    name="sample.jpg",
    file_path="./sample.jpg",
    annotations=annotations_map.get("sample.jpg")
)

Example of converting annotations to create multiple tasks.

In the case of the following tree structure.

dataset
├── sample1.jpg
├── sample1.json
├── sample2.jpg
└── sample2.json

Example source code.

import fastlabel

project = "YOUR_PROJECT_SLUG"
input_dataset_path = "./dataset/"

annotations_map = client.convert_labelme_to_fastlabel(folder_path=input_dataset_path)
for image_file_path in glob.iglob(os.path.join(input_dataset_path, "**/**.jpg"), recursive=True):
    time.sleep(1)
    name = image_file_path.replace(os.path.join(*[input_dataset_path, ""]), "")
    file_path = image_file_path
    annotations = annotations_map.get(name) if annotations_map.get(name) is not None else []
    task_id = client.create_image_task(
        project=project,
        name=name,
        file_path=file_path,
        annotations=annotations
    )

Please check const.COLOR_PALLETE for index colors.

Dataset

Create Dataset

Create a new dataset.

dataset = client.create_dataset(
    name="Japanese Dogs",
    slug="japanese-dogs",
    type="image"
)

Response Dataset

See API docs for details.

{
    'id': 'YOUR_DATASET_ID',
    'name': 'Japanese Dogs',
    'slug': 'japanese-dogs',
    'type': 'image',
    'createdAt': '2022-10-31T02:20:00.248Z',
    'updatedAt': '2022-10-31T02:20:00.248Z'
}

Find Dataset

Find a single dataset.

dataset = client.find_dataset(dataset_id="YOUR_DATASET_ID")

Success response is the same as when created.

Get Dataset

Get all datasets in the workspace. (Up to 1000 tasks)

datasets = client.get_datasets()

The success response is the same as when created, but it is an array.

You can filter by type and keywords.

datasets = client.get_datasets(
    type="image", # 'image', 'video', 'audio'
    keyword="dog"
)

If you wish to retrieve more than 1000 data sets, please refer to the Task sample code.

Update Dataset

Update a single dataset.

dataset = client.update_dataset(
    dataset_id="YOUR_DATASET_ID", name="World dogs"
)

Success response is the same as when created.

Delete Dataset

Delete a single dataset.

⚠️ The dataset object and its associated tasks that dataset has will also be deleted, so check carefully before executing.

client.delete_dataset(dataset_id="YOUR_DATASET_ID")

Create Dataset Object

Create object in the dataset.

The types of objects that can be created are "image", "video", and "audio". There are type-specific methods. but they can be used in the same way.

dataset_object = client.create_image_dataset_object(
    dataset_id="YOUR_DATASET_ID",
    name="brushwood_dog.jpg",
    file_path="./brushwood_dog.jpg",
)

Response Dataset Object

See API docs for details.

{
    'id': 'YOUR_DATASET_OBJECT_ID',
    'name': 'brushwood_dog.jpg',
    'size': 6717,
    'height': 225,
    'width': 225,
    'groupId': None,
    'createdAt': '2022-10-30T08:32:20.748Z',
    'updatedAt': '2022-10-30T08:32:20.748Z'
}

Find Dataset Object

Find a single dataset object.

dataset_object = client.find_dataset_object(
    dataset_object_id="YOUR_DATASET_OBJECT_ID"
)

Success response is the same as when created.

Get Dataset Object

Get all dataset object in the dataset. (Up to 1000 tasks)

dataset_objects = client.get_dataset_objects(dataset_id="YOUR_DATASET_ID")

The success response is the same as when created, but it is an array.

You can filter by keywords.

dataset_objects = client.get_dataset_objects(
    dataset_id="YOUR_DATASET_ID", keyword="dog"
)

If you wish to retrieve more than 1000 data sets, please refer to the Task sample code.

Delete Dataset Object

Delete a multi dataset objects.

⚠️ Related tasks will also be deleted, so please check them carefully before execution.

client.delete_dataset_objects(
    dataset_id="YOUR_DATASET_ID",
    dataset_object_ids=[
        "YOUR_DATASET_OBJECT_ID_1",
        "YOUR_DATASET_OBJECT_ID_2",
    ],
)

Get Import Histories For Dataset Object

Get all import histories in the dataset. (Up to 1000 tasks)

datasets = client.get_dataset_object_import_histories(
    dataset_id="YOUR_DATASET_ID"
)

Response Dataset Object Import Histories

See API docs for details.

[
    {
        'id': 'YOUR_DATASET_OBJECT_IMPORT_HISTORY_ID',
        'type': 'local',
        'status': 'completed',
        'msgCode': 'none',
        'msgLevel': 'none',
        'userName': 'admin',
        'count': 1,
        'createdAt': '2022-10-30T08:31:31.588Z',
        'updatedAt': '2022-11-02T07:36:07.636Z'
  }
]

API Docs

Check this for further information.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

fastlabel-0.11.28.tar.gz (49.5 kB view hashes)

Uploaded source

Built Distribution

fastlabel-0.11.28-py3-none-any.whl (31.9 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page