We Streamline AI Development
Collecting and annotating training data for autonomous systems is a very costly and labor intensive process. Moreover, those systems need to be tested and verified rigorously under a huge variety of scenarios and conditions before they can safely be deployed in the field.
We at RealSynth aim at minimizing the need of field data and field tests. We build fully tagged, realistic virtual environments to train and test AI faster, cheaper, and safer than the current manual process.Read More
What RealSynth stands for
Our synthetic environments and sensor models are designed for highest technically possible fidelity.
We emphasize seamless integration to existing development workflows through a cloud interface.
Run just a single, or hundreds of sessions in parallel. Drive billions of virtual test kilometers at the push of a button.
We support common data formats (KITTI, COCO, VOC) and interface with standard middleware solutions, such as ROS.
Push your algorithms to the limit and repeat critical, yet seldom edge cases over and over.
Control a large variety of parameters including sensor parameters (such as sensor/lens settings), lighting conditions, weather conditions, and scenarios.
RealSynth Under the Hood
A highly customized version of the Unreal Engine 4 enables fully automated capturing of ground truth data, rendering scenarios at different times of day, and various weather conditions in real-time.
Our technology allows providing enormous quantities of annotated image data fast, fully customized and at low costs.
Beyond that, our proprietary simulation environment is optimized for low latency, scalable cloud deployment.
The minds behind RealSynth
We are constantly scouting for outstanding talent to join our growing team. Please check AngelList for open positions or contact us directly to learn about opportunities.
A look at our future and past
Investors and Business Angels are welcome to our seed round!
In our first paid pilot project with a large corporate, we were able to show that blending our synthetic data with an existing real dataset can improve object detection performance significantly.
We successfully graduated from the TechStarts Internet of Things startup accelerator program in New York City.
RealSynth GmbH was founded as a spin-off from Siemens Mobility Technology & Innovation.
Don't hesitate to contact us
Interested in a pilot project or an investment opportunity?