Pre-annotation can help speed up the labeling process for various use cases and tasks.
You can use machine learning to speed up your Kili annotation project by:
- Importing labels
- Orchestrating active learning with an AutoML framework
- Implementing active learning strategies with queue prioritization
Some of our customers who implemented model-based pre-annotation managed to achieve impressive results. Here are some examples:
- Semantic segmentation: performance increased by 70% (medical imaging project).
- Bounding box detection: performance increased by 45% (facilities inspection project).
- NER and text classification: performance increased by 30% (bank and insurance project).
- Video object tracking: performance increased by 50%.
For an end-to-end example of how to programmatically import model-based pre-annotations to a Kili project using Kili's Python SDK, refer to our tutorial on importing assets and labels.
Updated 5 days ago