The ARTIEYE DMS Technology Suite enables fast and efficient development of fully customized computer vision-based AI DMS systems. Fully developed by Xylon, the ARTIEYE DMS suite includes HW/SW intellectual property, developed test data sets, biomedically established ground truth sets of data, data collection procedures for simulation lab and test fleets on open roads, Machine Learning (ML) methods, optimization of neural networks, hardware equipment for advanced Hardware-in-the-Loop (HIL) simulations with synthetically generated and road recorded drivers, as well as embedded hardware implementations.

ARTIEYE users may license, or even buy out the complete technology suite or parts of it. Besides the existing off-the-shelf parts, Xylon offers design services for system tuning and other flexible cooperation models.

License or own the source code, skip years of internal developments and take full control over your sensitive products!

Being AI-based, the existing ARTIEYE suite can be organically expanded to support other in-cabin applications, such as occupancy monitoring, hands on-wheel and so on.

 

Embedded AI Implementation

  • AMD Zynq™ UltraScale+ based implementation. Targeting small and cost-effective chips. Works in Xylon’s demo camera and runs on an off-the-shelf evaluation board (ZCU104)
  • Embedded firmware consists of IP cores for programmable SoC and software
  • Main modules: drowsiness detector and detectors for visual, cognitive and manual distraction
  • Supporting modules: Head and Eyes detection, Face Landmarks, Gaze tracking, NIR ISP pipeline, control and alarm module…
  • Possible porting to different embedded platforms
Read more

Machine Learning and Neural Networks Optimization

  • Post-processing of acquired road data – metadata extraction
  • High-performance GPU-based infrastructure setup and coding
  • Extraction of essential metadata for automated learning
  • Neural networks pruning for best cost/performance ratio

Labs setup

  • Driving simulation lab with multiple Near-infrared (NIR) cameras, multi-channel biomedical monitoring cameras and reference monitoring cameras
  • Driving scenarios and simulation protocols development
  • Gaze tracking lab with the golden reference third-party system included
  • NIR Image Signal Processing (ISP) lab for driver monitoring camera tuning
Watch the video

Data Acquisition, Pre- and Post-Processing

  • Protocols for selection of research participants
  • Build-up of computer vision datasets
  • Data collection within a simulation environment, closed testing tracks and open road environment
  • Programming of synthetically generated driver models
  • Use of Xylon data loggers, equipping of simulators and test vehicles
Read more