Data labeling automated with 10 annotation

LABELING AI is a deep learning-based Auto-Labeling technology that utilizes AI to automate and quickly label large amounts of data, after labeling only a small amount of data manually.

Easier data labeling with just inspection

Auto-Labeling is performed with artificial intelligence and a large amount of data can be labeled only by inspecting the results.

The most efficient way to label your data

Deep learning-based Auto-Labeling can be performed with only 10 labels, and a large amount of data can be labeled only by inspecting and correcting the result of Auto-Labeling.

1. Labeling Manually

Manually generate 10 labeled data.

2. Training Model

Train an auto labeling AI with the 10 pre-labeled data. Review and correct the results to enhance auto labeling performance.

3. Deploy the best AI

Repeat the previous step to generate 1,000, 10,000, or 100,000 auto labeled data. Transform your auto labeling AI into an object detection AI model to perform object detection as needed.

Convenient labeling inspection and modification with smart magic tool

Labeling inspection and modification are also easily performed through the magic tool that can quickly label even complex-shaped objects.

We offer a variety of smart labeling tools.

We provide a variety of labeling tools such as Bounding Box, Polyline, and Polygon magic tool.

Semantic Segmentation

It recognizes and classifies the entire image by region, and is actively used in autonomous driving technology.

Skeleton

It recognizes and tracks human motion by detecting the skeleton, and is used in motion detection fields such as sports or security.

Free data import/export

Regardless of the labeling work environment, various coordinate data can be imported to LABELING AI to perform labeling, and the processed labeling coordinate data can be extracted in various formats(COCO JSON / VOC).

COCO JSON
{
    "images": [
        {
            "id": "60a212aac869a1fea276480d",
            "file_name": "/images/img_labelingExample.jpg",
            "width": 4000,
            "height": 2084
        }
    ],
    "type": "instances",
    "annotations": [
        {
            "segmentation": [
                [
                    1200,
                    907,
                    1200,
                    1882,
                    2903,
                    1882,
                    2903,
                    907
                ]
            ],
            "area": 1660425,
            "iscrowd": 0,
            "ignore": 0,
            "image_id": "60a212aac869a1fea276480d",
            "bbox": [
                1200,
                907,
                1703,
                975
            ],
            "category_id": 2621,
            "id": "60a216ae2cd9eb1bbde44e2b"
        }
    ],
    "categories": [
        {
            "supercategory": "none",
            "id": 2620,
            "name": "bus"
        },
        {
            "supercategory": "none",
            "id": 2621,
            "name": "car"
        },
        {
            "supercategory": "none",
            "id": 2622,
            "name": "truck"
        }
    ]
}

VOC
<annotation>
  <folder>VOC2007</folder>
  <filename>img_labelingExample.jpg</filename>
  <source>
    <database>The VOC2007 Database</database>
    <annotation>PASCAL VOC2007</annotation>
    <image>none</image>
    <flickrid>none</flickrid>
  </source>
  <owner>
    <flickrid>none</flickrid>
    <name>[email protected]</name>
  </owner>
  <size>
    <width>4000</width>
    <height>2084</height>
    <depth>3</depth>
  </size>
  <segmented>0</segmented>
  <object>
    <name>car</name>
    <pose>Unspecified</pose>
    <truncated>0</truncated>
    <difficult>0</difficult>
    <bndbox>
      <xmin>1201</xmin>
      <ymin>908</ymin>
      <xmax>2903</xmax>
      <ymax>1882</ymax>
    </bndbox>
  </object>
</annotation>

80% cost reduction

The entire process from training data preparation to AI deployment can be solved with LABELING AI, saving the cost and duration of the project.

Traditional labeling
LABELING AI

Artificial intelligence development cost

Labeling inspection cost

Manual labeling cost

Labeling tool cost

* Only the auto-labeling results 

that passes the inspection

 will be processed for payment.

Auto-Labeling for continuous training

Starting with the first 10 manual labeling, the performance of LABELING AI is further improved by repeating the inspection/correction and re-training of Auto-Labeling results.

Project management through dashboard

Check and manage work status at a glance through the dashboard for the labeling project.

  • ✓ Data labeling progress
  • ✓ Status by labeling class
  • ✓ Work status of shared workers

Auto-Labeling using various artificial intelligence

LABELING AI supports various methods to perform Auto-Labeling.

  • ✓ Custom AI: Perform Auto-Labeling by training manually labeled data
  • ✓ General AI: Auto-Labeling immediately without training
from ds2 import DS2

ds2 = DS2(apptoken=“s2234k3b4”)
ds2.startAutoLabeling(
    "people.zip",
    "1000",
    has_label_data=True,
    ai_type="general",
    autolabeling_type="box",
    general_ai_type="person",
    model_id=None,
    custom_ai_stage=0,
    workapp="object_detection"
)

SDK support for convenient programming development

Build a data pipeline for MLOps at DS2.ai with Python. SDK gives you access to all of the processes from uploading and labeling data, to training and deploying the artificial intelligence model.

Learn more →

Auto-Labeling for continuous training

Starting with the first 10 manual labeling, the performance of LABELING AI is further improved by repeating the inspection/correction and re-training of Auto-Labeling results.

Project management through dashboard

Check and manage work status at a glance through the dashboard for the labeling project.

  • ✓ Data labeling progress
  • ✓ Status by labeling class
  • ✓ Work status of shared workers

Auto-Labeling using various artificial intelligence

LABELING AI supports various methods to perform Auto-Labeling.

  • ✓ Custom AI: Perform Auto-Labeling by training manually labeled data
  • ✓ General AI: Auto-Labeling immediately without training

SDK support for convenient programming development

Build a data pipeline for MLOps at DS2.ai with Python. SDK gives you access to all of the processes from uploading and labeling data, to training and deploying the artificial intelligence model.

Learn more →
from ds2 import DS2

ds2 = DS2(apptoken=“s2234k3b4”)
ds2.startAutoLabeling(
    "people.zip",
    "1000",
    has_label_data=True,
    ai_type="general",
    autolabeling_type="box",
    general_ai_type="person",
    model_id=None,
    custom_ai_stage=0,
    workapp="object_detection"
)

Get started right away with the most efficient labeling method.