Kili app manages a queue for each project. The queue can be customized and redefined throughout the project. This is helpful for advanced annotation processes for business expectations, priorities, active learning, production supervision, and real-time annotation.
Three strategies are possible:
FiFo is the default queue orchestration. Assets are loaded based on their time stamp (creation date): assets uploaded as the first ones will be distributed to be annotated first.
You can customize the queue to assign a specific list of assets to specific labelers.
The assignments can work for the whole queue or for a specific subset of the dataset.
Check the details in Kili API references:
- GraphQL API key:
- Python SDK, function:
For examples, refer to this recipe:
Additionally, you can assign assets to specific labelers through the UI. For details, refer to the Queue page description.
Prioritizing assets allows you to set the specific priority at the asset level. You can either set a priority to each asset or set a priority to a limited number of assets within the dataset. The higher the number, the higher the priority. If two assets have the same priority, FiFo will apply.
- If asset 1 has priority N_1 and asset 2 has priority N_2, and N_1 > N_2, then asset 1 has higher priority than asset 2 and will be distributed first.
- If N_1 = N_2, priority follows the default strategy (FiFo).
Also, refer to this example recipe:
Additionally, you can set priority on specific assets through the UI. For details, refer to the Queue page description.
You can also prioritize assets by using Kili AutoML.
For an end-to-end example of how to programmatically prioritize your project queue and assign specific assets to specific labelers using Kili's Python SDK, refer to our tutorial on how to set up workflows.
Updated 5 days ago