The open source active learning framework for computer vision

A test suite for your labels, data, and models. Encord Active helps you find failure modes in your models, prioritize high-value data for labelling, and drive smart data curation to improve model accuracy.

Request early accessRead the docsJoin our Slack community
  • 🚀 The fastest way to debug your dataset and improve model performance
  • 🤠 Built for ML Engineers and Data Scientists
  • One line of code to get started to import your labels and model predictions
$ pip install encord-active
$ encord-active --help
How it works

Enhance data quality and model visibility

Encord Active has been designed as a one-stop shop for data quality and model debugging. We enable world-leading computer vision teams to rapidly improve their datasets and models.

Data Quality

The computer vision-first data quality toolkit

Discover errors and outliers within your data - all in one platform.

Uses platform

Understand your data distribution

Encord Active enables you to explore your data distribution with respect to a set of customisable metrics. We index your data by different metrics to uncover areas where your data is underrepresented or mislabelled.

Uses platform

Find outliers

With Encord Active, you can quickly find image outliers with respect to a set of customisable metrics. Encord Active finds outliers using precomputed interquartile ranges.

Uses platform

Find similar images

Encord Active's similarity search means you can easily find semantically similar images in your dataset. Identifying these cases can help you understand where your data is underrepresented or mislabelled.

Label Quality

Ensure your annotations meet your quality standards

Discover hidden issues with your annotations. Save time and improve label quality by reducing manual work and augment your experts with AI.

Png feature model import

Understand your label distribution

Encord Active enables you to explore your label distribution by pre-defined metrics, custom metrics, and label classes. Understanding your label distribution by different metrics can help you discover gaps in your dataset to prioritise for further data collection and labelling

Downloading models

Find outliers in your labels

With Encord Active, you can quickly find image outliers for pre-defined metrics, custom metrics, and label classes. Encord Active finds outliers using precomputed interquartile ranges.

Downloading models

Explore annotator performance

Discover biases and performance issues in your annotation process and workforce.

Model Quality

Gain visibility into model performance

Encord offers intelligent model evaluation features, ensuring that your models are of the highest possible quality before they go into production.

Quality prediction iou grid in program

Understand model performance

Use Encord Active to understand where your model is performing well and where it is not. Encord Active provides a range of metrics to help you understand your model performance.

Png feature model import

Find what features matter most to your model

Encord Active correlates model performance to your data distribution to help you understand what data features drive model performance.

Png review json

Explore model false positives

Encord Active quickly allows to identify false positives by comparing your model predictions to your pre-existing labels.

Talk to sales