Camera based ADAS and AD Product Accelerators

Kick-start your camera based ADAS development with PathPartner’s camera based ADAS and AD product accelerators. PathPartner’s camera based ADAS and AD product accelerators are a set of algorithms designed to provide a solid foundation for developing camera-based ADAS and AD solutions. The growing set of algorithms currently includes components for traffic sign recognition, traffic light recognition, vulnerable road-user detection, road-lane markings and semantic segmentation. Available on major automotive platforms including; NVIDIA Tegra K1, X1, X2 and AGX Xavier, Texas Instruments’ TDA2x/3X, Renesas R-Car V3M/H, Qualcomm 820A, 625, 635 (In progress), NXP’s S32V234 (In progress), windows/Linux PCs and server grade platforms for quick evaluation.

Why PathPartner Camera ADAS Product Accelerators?

Simplify your ADAS solution evaluation and development

With readily available algorithms and sample applications, you don’t need to worry about implementation details

Quickly realize complex vision algorithm pipelines

Offer a range of algorithms and pipelines that can be easily customized for your requirement

Select a package available on platform of your choice

With availability on major automotive platforms, you are not tied to a single platform

Vulnerable Road User Detection and Classification

  • Custom 22-layer deep neural network based on SSD architecture, for detecting and classifying the vulnerable road users and road objects.
  • Detection network is optimized using network pruning and local re-training methods generating an optimal network that runs real time on embedded SoC’s

Traffic Sign Recognition

  • Custom 22-layer deep neural network based on SSD architecture, for detecting and classifying the traffic signs. Apart from detecting the position, the traffic sign boards are further classified into 43 specific (German) signs.
  • Detected traffic lights are further classified based on colour of the light (off, red, green and yellow). Detection network is optimized using network pruning and local re-training methods generating an optimal network that runs real time on embedded SoC’s

Traffic Light Classification

  • Custom 22-layer deep neural network based on SSD architecture, for detecting and classifying the traffic lights.
  • Detected traffic lights are further classified based on colour of the light (off, red, green and yellow). Detection network is optimized using network pruning and local re-training methods generating an optimal network that runs real time on embedded SoC’s

Driveable Road Area Detection

  • Computationally light weight machine learning method that runs on ARM CPU.
  • Detects and localizes lane markings on highway roads, in real time. Can detect curved lanes along with the straight lines.

Semantic Segmentation

  • Customized deep neural network based on Shuffle Net architecture to classify image pixels belonging to one of the trained classes.
  • Real time Pixel-wise Segmentation of Images into multiple road object classes. Segments the given frame as Road (Drivable region), Pedestrian, Bicycles/Motorcycles and Cars/Trucks/Bus classes.

By submitting this form, you authorize PathPartner to contact you with further information about our relevant content, products and services. You may unsubscribe any time. We are committed to your privacy. For more details, refer our Privacy Policy