Skip to content
Rendered.aiDocs
Application User Guides › Tutorials › Creating and Using Datasets

Dataset Annotations

Annotations are descriptions of imagery contents formatted for machine learning algorithms. Rendered.ai provides conversion services to multiple standard formats.

Supported Annotation Formats

The platform supports the following annotation formats:

FormatDescription
COCOCommon Objects in Context (COCO) dataset was created for object detection, segmentation and captioning
GEOCOCOGeospatial data annotation format compatible with COCO tools
PASCALPASCAL VOC challenge provides standardized datasets for comparing model performance
YOLOYou Only Look Once (YOLO) object detection system
KITTIFormat for image classification
SageMaker ODObject detection using MXNet
SageMaker SSFor use with semantic Segmentation Algorithm

Mapping Annotations

Users can map objects to specific classes to consolidate similar items. For example, you might group various toy types and cube variants into two categories using a YAML annotation map.

The annotation map is a YAML file with two sections:

  • classes: Integer keys paired with class lists (supporting hierarchies)
  • properties: Python eval statements determining class assignments based on object metadata

Here is an example annotation map:

yaml
classes:
    0: [none, Cubes]
    99: [none, Toys]
properties:
    obj['type'] == 'YoYo': 99
    obj['type'] == 'BubbleBottle': 99
    obj['type'] == 'Skateboard': 99
    obj['type'] == 'Cube': 0
    obj['type'] == 'Mix Cube': 0
    obj['type'] == 'PlayDough': 99

Creating Annotation Maps

To create an annotation map:

  1. Navigate to your organization's Annotation Map tab.
  2. Click the + New Annotation Map button.
  3. In the dialog, enter a name for the map and upload the YAML file. Optionally add a description.
  4. Click Create.
New Annotation Map
New Annotation Map

Creating Annotations

To create annotations for a dataset:

  1. Navigate to your workspace Resources and add the Annotation Map to the workspace.
  2. Select the dataset in the Dataset Library.
  3. Click the + icon next to Annotations in the right-side panel.
  4. Select the annotation format and annotation map from the lists.
  5. Click the Create button to start the annotation job.
Create Annotations
Create Annotations

Once the job is complete, you can download the annotation files via the download icon.

Job Status

Three status states are available:

  • No symbol: The service job is complete and ready to use.
  • Sand dial symbol: The job is running. It will remain this way until the job has either completed or failed.
  • Error symbol: The job has an issue. You can click on the symbol to fetch a service log.