Premiere of AI in German live-TV

In a premiere on German TV in the TV-show ” Bundespolizei Live – Großkontrolle an der Grenze” on 18.09.2019 by Spiegel TV and Kabel1 our software for live-pixelation of faces was used for the first time in production. The AI software provided by Evotegra GmbH processes 50 HD television pictures per second and detects and obscures all faces in the image. This enables us to meet the particularly high demands on the protection of personal rights in cooperation with the Federal Police. Due to Evotegra GmbH’s high-performance AI software platform, it was possible to realize the project in a mere 4 weeks from initial request to live-broadcast. Three servers, each with one GPU, processed approximately 1 gigabyte of data per second during the 2-hour transmission.
Many thanks to the great team of Kabel1, Spiegel TV and partners for the excellent cooperation and support.

See Evotegra Software in action in the live recording at: 30:0730:5131:3436:1136:331:06:00

Data collection system successfully tested

(14-08-2019) In cooperation with our partner Rebotnix our Smart-City Visual Inspection System was successfully tested in a Remondis truck. The system consisting of Rebotnix in-car system based on the NVIDIA Nano, camera and GPS successfully recorded data for a smart maintenance use-case for traffic signs.

Autonomous car  implemented and tested

(10-07-2019) Evotegra successfully implemented and tested a mapless autonomous car. All relevant software and hardware was developed in-house and the costs for hardware excluding the car fits into a $10k budget. Impossible? See yourself.

Gutenberg Digital HubGutenberg Digital Hub

AI in Depth

(02-07-2019) In cooperation with our partner Becom we just published our experience with Time of Flight (ToF) cameras in combination with Deep Learning in the well known InVision magazine. Depth adds more entropy compared to color in most use-cases and hence deep learning has the potential to achieve higher recognition rates compared to common RGB images. 3D sensors with synchronous images in combination with Deep Learning will play a major role in many industrie use-cases in the coming years.

Read the article here.

Evotegra joins NVIDIA’s Inception Program

Mainz, 13-06-2019. Evotegra has joined the NVIDIA Inception program, which is designed to nurture startups revolutionizing industries with advancements in AI and data sciences.

The NVIDIA Inception program helps Evotegra to get in touch with technical expert and resources to get the best performance for our solutions.

NVIDIA’s Inception program is a virtual accelerator program that helps startups during critical stages of product development, prototyping and deployment.

Gutenberg Digital Hub

Evotegra Data Collection System announced

Evotegra GmbH is glad to announce a professional full stack in-vehicle data collection solution starting in Q3 2019. The system will include all hardware and software to collect data on the spot. The deep learning enabled system includes single or stereo camera operated by a Nvidia based embedded system with 10-50W power consumption, localization and a cloud based data collection back-end. Additional services include SLAM, path planning, actuation, highly automated data collection and labeling as well as training of custom neural networks. The system can be highly customized to fit any use case ranging from high end for automotive to large scale fleet deployment. For the low end system in high volumes the target price is less than 1000 EUR.

Download our Evotegra Stereo Camera Fact Sheet to get an impression of what you can expect from a Stereo Vision System.

Evotegra goes Nano


We are proud to show our Stereo 3D-Reconstruction & Deep Learning Object Detection solution on the Nvidia Jetson Nano.

Technical details:

  • Nvidia Jetson Nano
  • 2 x Basler acA40-uc
  • Resolution: 1900×1200 pixels
  • Power consumption including peripherals: ca. 15-18W
We achieve the following performance:
  • Deep Learning based object detection on one camera without visualization: 10 images per second
  • 3D reconstruction with stereo-camera: 9 images per second
  • Stereo 3D-Reconstruktion with simultaneous Deep Learning based Object Detection including 3D visualization: 5 images per second
Gutenberg Digital Hub

In-Car Stereovision & Deep Learning

Check out our latest in-car demo of 3D Stereo-Reconstruction and simultaneous Deep Learning Object Detection.

Some technical facts:

  • Platform: Nvidia Xavier
  • Camera: 2x Basler acA 40-uc
  • Resolution: 1920×1200 Pixels
  • Data: 20 images or 1.5GBit (180 Megabytes) per second
  • Power consumption: ~45W including all peripherals
  • 3D Points: 2 up to 10 Millions per second
  • Visualization: 3D and Object Detection