Understand the
road ahead

understand.ai provides high-quality training and validation data to enable mobility companies to develop with confidence computer vision and machine learning models that reliably and safely power autonomous vehicles.

Request a free demo

Ground-Truth quality
annotations

Achieve the most reliable safety standard and the best results by training and validating your algorithms with accurate annotations made with German precision.

Fast and flexible
scalability

Our system is fast and flexible in its approach. Manage projects with more than 100 million annotated images up to 5 times faster than conventional approaches.


Compliant and secure
data ownership

Own your data throughout the annotation process. Our processing is compliant to the newest GDPR regulations via services like advanced data anonymization and de-personalization. For additional privacy and security, EU-based data centers are available on request.

Team of technical
experts

Profit from our creative solutions, which are adapted and tailored to your project with its unique requirements and success criteria. We will make sure to meet your individual timeline.

Training and Validation Data

Our portfolio covers the diversity of all regular raw data formats, project requirements and annotation types. This includes 2D images, semantic enhancements, video annotations and LiDAR.

Bounding Box
Original

Bounding Boxes

The annotation type with the most scientific research and most commonly used is bounding boxes. They are easy to apply to machine learning models and faster to annotate in comparison to other annotation types.

Unlike segmentation, bounding boxes may also contain invisible parts of the classified object by approximating occlusions. Due to the inherent instance-awareness of bounding boxes, your algorithms will get a better understanding of the concept of specific objects.

Segmentation
Original

Pixelwise Segmentations

Since the world is not made out of boxes, we are also offering a more precise method to annotate your data - semantic segmentation.

Depending on the raw data, bounding boxes can contain noise in the form of background and occlusions. This is tackled with semantic segmentation, where each pixel assigned to the class of your selected objects will be annotated. It is therefore the closest to a true representation of reality in 2D space, regarding class assignments. It also is more versatile, since it is very easy to distinguish between objects, e.g., road, lanes and curbs and it is possible to annotate classes that are not instantiable.

Object Video Tracking

The annotation type video annotation has advantages compared to still-frame bounding boxes, segmentations or other ways of labeling. Rather than multiple different images, it is a consistent sequence of images of one particular scene resulting in a more informative context for the algorithms.

In that way an object gets the same attributes across multiple frames. It’s important to track moving objects in a scene to be able to learn about the object's behavior, intent and trajectory. By learning these attributes over time, vehicles will be able to anticipate and cope with rational and even irrational human behavior.

Raw video data source: Garmin Eesti

Bounding Cuboids

3D Boxes in LiDAR
Point Clouds

LiDAR is an active optical sensor technology which scans the earth’s surface to determine highly accurate x, y and z measurements. It transmits laser beams to a specific object and reflects its movements back to the receiver and analyzes the time span and distance with GPS and INS information to construct a 3D point cloud of reflective obstacles.

With this raw data we are able to identify objects in a 3D point cloud by annotating 3D bounding cuboids around the specified object.

Anonymized
Original

Identity Protection Anonymizer

With GDPR autonomous vehicles need to be able to collect street scene data with all the key personal information automatically removed.

Our new AI-powered anonymizer tool, called Identity Protection Anonymizer, ensures your data is compliant by blurring faces and number plates in a fully automated fashion. Our system can be used on premise or as a service via the understand.ai Cloud Platform. In both cases we make sure the anonymizer provides the highest quality data at the scale required.

understand + ai

Our mission is to push the boundaries of technology to advance the state of
autonomous driving.

UAI's method of annotation

Traditional Manual Labeling

Our company was founded in the belief that a human’s ability to view and interpret an object in a particular way, can be complemented, enhanced and accelerated by applying artificial intelligence to a highly repetitive task.

understand.ai provides high-quality training and validation data to enable mobility companies to develop with confidence computer vision and machine learning models that reliably and safely power autonomous vehicles. The opportunity to enable new modes of mobility and give consumers increased choice inspires us everyday.

Our advanced capabilities include bounding box annotations and semantic segmentation for 2D camera data as well as LiDAR footage, with specific meta-attributes and instance ID labeling for object tracking.

By applying specialized artificial intelligence technology to repetitive tasks and quality check every single image with our in-house multi-level quality assurance team, we are able to quickly and precisely annotate data and thus accelerate the production of the ground-truth data required to make autonomous driving a reality.

understand.ai is headquartered in Karlsruhe, Germany and has offices in Berlin and San Francisco. Our engineers have relevant experience gained at innovative companies such as BMW, Google and Mercedes-Benz. Contact us today to better understand the road ahead.

Work with us

Want to turn science-fiction into reality and make self-driving cars become part of our daily life while also working and having fun with super awesome people? We’re hiring. If you belong at the forefront of technology shaping the future, enjoy collaborating with curious and creative people, and believe you make a difference, apply today.

See open positions

Geeks talk mobility

Passionate about accelerating the state of the art in autonomous driving, the ability of machine learning to improve reliability and safety, advanced approaches to the annotation of vast amounts of data? You get the idea. Head over to our blog, Geeks Talk Mobility, and join the discussion.

Thank you!

Your message has been successfully sent. We'll be getting in
touch with you soon!

Contact us

Launch your next autonomous driving project with high quality training and validation data. Contact us today on the ideal solution that meets your needs.

We're sorry, but something went wrong trying to transfer your message. You can either try again or write us an email directly to [email protected].

Our offices
Karlsruhe (HQ)

Hirschstr. 71
76133 Karlsruhe
Germany

Berlin

Rheinsberger Str. 76/77
10115 Berlin
Germany

San Fransisco

1355 Market Street
94103 San Francisco
USA