Perception analytics

Use powerful and efficient analytics tooling to track and measure key performance indicators. Measure the quality of your data set and connect your sensor output directly to ground truth data to validate essential performance and safety requirements.

arrow down icon

Get detailed quality and distribution statistics of your perception data

Perception analytics tooling

Screenshot of perception analytics from the Annotell platformMacBook frame image

Quantify quality

Analyze and evaluate your datasets with help of detailed quality and distribution metrics. How certain are you that your ground-truth is correct? Don't take it for granted, quantify it.

Validate with ground truth

Use the powerful perception analytics tooling to compare sensor output directly against labeled ground truth data and validate that your sensors meet important KPIs.

Did you know?

Knowing important statistics about precision, recall and accuracy enables you to know more exactly why your model performs as it does, giving you the advantage to specifically target and increase certain aspects of your model.

Validate that your safety-critical sensors meets performance

Direct sensor integration towards ground truth

Connect and automatically re-simulate your sensor output directly towards ground truth data to ensure that your sensor meets it’s performance goals. With Annotell’s KPI SDK you can run large scale KPI calculations over thousands of kilometers of collected ground truth data. Schedule and automate reports to track your sensor performance after new software releases and communicate your progress to relevant stakeholders.

Illustration of cars with sensors

Verify objective safety with statistical certainty

Fully customizable quality metrics and reporting

The Perception Analytics tooling enables you to craft, analyse and evaluate highly important metrics of your output with statistical measures of certainty to ensure that your sensors meets it’s requirements.

Image of a quality report

When dealing with large validation projects, it’s common that the data volumes are in the regions of PetaBytes of data. If not handled in a smart way, moving this amount of data can easily go up to the 100 000€ range.

Schedule a demo to understand how we can support your specific use case