CarEye® Perception Software
Perception-Software for ADAS & AD
The CarEye® Perception Software is a tool for developing collision detection systems by using the advanced capabilities of self-trained, specialized EYYESNET neural network and the modular EYYES software framework. Custom ADAS functions can be configured very quickly via the graphical user interface, from simple reversing assistants to complete vehicle solutions with any number of cameras. Various included high-end features help to optimize the detection quality and perfecting the final ADAS function.
Main properties
- Available for different hardware platforms
(i.e. NVIDIA, NXP, TDA, Xylinx ultra scale) - Different camera input tools (IP, FPD Link III, GMSL2)
- Graphical user interface with customizable user rights
- Integration via Docker image
- EYYESNET self trained CNNs for typical traffic objects (Pedestrians, cyclists, cars, trucks, trams)
- General purpose CNN or specialized DNNs for different camera perspectives (mounting position, FOVs, classes, environmental conditions)
Features
- Object detection, object classification, object tracking
- 2D/3D estimation
- Localization of objects
- Movement prediction
- 3 different warning strategy approaches available
- EYYESNET or customer neural networks
- Parallel processing of different neural networks
- Overlays like bounding boxes, etc.
- System self-monitoring
- Lens contamination monitoring
- High level logging and debugging possibilities
- Easy adaption of the processing pipeline
- Video-output H264/H265 stream or HDMI
- Object List Streaming via UDP
For more information or to request a consultation, please contact us!