Open World Object Detection

The Open World Object Detection plugin allows you to detect a great variety of objects on image files in your project.

The plugin will draw a bounding box around each object type it can find, and it will write the type of object in a nested text classification in the box itself.

Plugin Functionality

Adding the plugin to your organization will allow you, with one click, to detect a variety of objects on image (.jpg, .png) assets in the project of your choice. The plugin will draw a box around each object it can detect, and it will add the name of the class it detected to a text classification nested into the bounding box itself.

The plugin will create bounding boxes not belonging to any class.

The following are the classes of objects this plugin will detect:

10KB

class_names.txt

Using the Plugin

From the Plugin Directory, search for the name of the plugin and add it to your organization. More information on installing plugins can be found in the Installing Plugins page.

From the Project Settings

Navigate to the project where you'd like to perform the object detection task.

Enter the Settings tab, then the Plugins section.

Find the plugin, and click on Open. A dialog will appear:

Ensure the dot before the plugin's name is green. The dot indicates the plugin's status, where green signifies the plugin's code is running, and red signifies the plugin's code is not running.

You will not be able to run plugins the code of which isn't running.

Do the following:

1. In the "Asset ID" field, enter the ID of the asset you would like to annotate.

2. In the "Task ID" field, enter the ID of the task you would like to annotate. (you may find info for steps 5 and 6 by entering the task and opening the "Task Info" drawer on the right-hand side, or by obtaining project information from the SDK.)

Click on Run. The plugin will start detecting objects in the image.

From Workflow

In the project workflow, add a Plugin-type stage such that it receives the assets you wish to annotate as input.

Click on the plugin-type stage and select the plugin, then alter its settings as necessary:

Click on Save to save the workflow. The plugin will now run every time it receives a compatible image asset as input.

From the Labeling Editor

With the image labeling editor open, click on the "Plugin Dropdown" to see a list of available model plugins:

If a default configuration was set, click directly on the plugin to run it immediately with the pre-set configuration. Otherwise, click on the three dots next to it to open the Run menu.

From the Run menu, configure its settings and click on Run to run it on the current image.

Plugin Configuration Settings

{
  "threshold": 0.5
}