Toloka documentation

Object recognition and area selection

For this type of project, you can use the Object recognition & detection preset.

This preset is designed to outline objects in images for training computer vision to detect them.

Take a look at the example: the labeling interface includes an image with bounding boxes, polygons, or key points. Note that validation, keyboard shortcuts, and task layout are already configured in this example.

Components used in the example
  • field.image-annotation: Lets you select areas using points, rectangles, and polygons.

    Show code
    {
      "type": "field.image-annotation",
      "image": {
        "type": "data.input",
        "path": "image"
      },
      "fullHeight": true,
      "data": {
        "type": "data.output",
        "path": "result"
      },
      "validation": {
        "type": "condition.required",
        "hint": "Please select an area"
      }
    }
    
  • condition.required: Checks if at least one area is selected.

    Show code
    {
      "type": "condition.required",
      "hint": "Please select an area"
    }
    
  • plugin.toloka: Customizes the task layout.

    Show code
    {
      "type": "plugin.toloka",
      "layout": {
        "kind": "pager"
      }
    }
    
  • plugin.field.image-annotation.hotkeys: Sets hotkeys to select area types and selection modes and to confirm or cancel area creation.

    Show code
    {
      "type": "plugin.field.image-annotation.hotkeys",
      "labels": [
        "1",
        "2",
        "3",
        "4",
        "5"
      ],
      "modes": {
        "select": "q",
        "point": "w",
        "rectangle": "e",
        "polygon": "r"
      },
      "confirm": "a",
      "cancel": "s"
    }
    

Note

You can add media files (audio files, videos, images) from your own server, Yandex Disk, or a cloud storage like Yandex Cloud, Google Cloud, or Amazon AWS.

If this preset doesn't meet your needs, see other examples in this section.

Add a description

To add a detailed description to the task, use the label property of the field.image-annotation component.

Show code
{
  "type": "field.image-annotation",
  "image": {
    "type": "data.input",
    "path": "image"
  },
  "fullHeight": true,
  "label": "Look at the picture and outine the road signs",
  "data": {
    "type": "data.output",
    "path": "result"
  },
  "validation": {
    "type": "condition.required",
    "hint": "Please select an area"
  }
}

Add a response field

To let Tolokers leave comments about the task or their response, add a text field using field.textarea. You can place it in the view.list along with the field.image-annotation component.

Show code
{
  "type": "field.textarea",
  "label": "Comments",
  "placeholder": "Enter text",
  "data": {
    "type": "data.output",
    "path": "comment"
  }
}

Add a layout

To enhance Toloker's experience, you can highlight different types of data with colors using view.alert. You can place it in the view.list along with the field.image-annotation component.

In this example, the text is highlighted with a yellow border.

Show code
{
  "type": "view.alert",
  "theme": "warning",
  "content": {
    "type": "view.text",
    "content": "Look at the picture and outline the road signs"
  }
}