PathPartner Technology

India & USA

PathPartner is a product R&D and engineering specialist. We help our clients develop, productize, and maintain advanced technology products. With our full stack engineering services, re-usable solution accelerators, and an unparalleled experience in transforming innovative ideas into full-fledged products, PathPartner provides its clients the advantage of top-of-the-line technologies, superior performance and faster time-to-market. With extensive expertise in embedded systems, IoT, computer vision, deep learning, multimedia and imaging technologies, PathPartner provides next-gen technology solutions and services to customers.

Services

Video Codecs

HEVC Encoder on Xeon

  • 422, 10-bit
  • 4K30 resolution / frame-rate
  • Memory Optimized
  • High Quality solution, better than Intel solution

 

HEVC Decoder on FPGA

  • 422, 10-bit
  • 4K60, 1080p120
  • Memory Optimized
  • Resource Optimized
  • Fully compliant universal decoder
  • Available also as SDAccel solution running as FFMPEG Plugin on Amazon EC2 as F1 instance

 

H.264 Decoder

  • Compliant with the standard ISO/IEC 14496-10 or ITU-T H.264
  • Supports Baseline profile and Main profile
  • Supports up to level 3.1, 1080p @ 30 fps
  • Supports robust error concealment
  • Easy to integrate APIs

Audio Codecs

AAC-LC encoder has been optimized for ARMv7, ARMv7-A and ARMv7E-M architectures

 

  • Supports fixed-point
  • Supports sampling rates of 8 KHz (narrow band) to 48 KHz (full band)
  • Supports bit-rates: 32 kb/s to 320 kb/s
  • Supports constant bit-rate (CBR), variable bit-rate (VBR)
  • Supports both mono and stereo coding
  • Supported Tools: TNS & M/S stereo pre-processing

Driver Monitoring System

Hardware plus software licensable solution

 

Advanced facial analysis algorithms

 

Deep learning models to assess driver’s alertness and focus under challenging environmental conditions

 

Object Detection

 

Face Detection

  • Detects driver drowsiness
  • Detects driver distractions

 

Face Recognition

Reduces & authenticate driver

Imaging Algorithms

Licensable solution

  • Sensor independent implementation
  • Supports both Bayer and YUV format.
  • No detail loss and no colour and edge artefacts due to enhancement.
  • Automatic colour enhancement
  • 2D Noise Reduction without detail loss

 

  • Efficient exposure fusion technique from multiple exposure images retaining best information from each of the images
  • Efficient image registration and de-ghosting algorithms
  • Can be easily ported to custom hardware solutions with HAL and camera driver access

 

  • Support for traditional RGGB Bayer sensors and RCCC (Red/Clear sensors)
  • Edge aware CFA Interpolation
  • 2 Noise filter modules – for RAW and YUV
  • Supports Auto white balance, auto exposure
  • Available on NVidia Jetson TK1 and easily portable on custom hardware solutions

Camera Modules

Stereo Camera will be designed using two OV10823 1/2.6” 10.5Mp Digital Image Sensor from Ominivision

Output formats supported by OV10823 sensor is 10-bit Raw RGB data and DPCM 10-8 Compression.

Board Size is 100 mm x 40 mm

Input Power to the device is 3.3V/5V power is received through MIPI connectors.

 

 

 

 

UVC and UAC Compliant Plug and Play device

Platforms: Linux, Windows, Jetson TX1/TX2

0 interface

  • It is backward compatible with USB 2.0 if USB3.0 is not available
  • Output Format: YUV2

 

 

UVC Compliant Plug and Play

0 interface

Output format supported

  • Monochrome camera (PPMVM-AP- 1)-Bayer Raw
  • Color camera (PPMVR-AP- 1)- Bayer Raw

AR0134 CMOS monochrome and colour sensor

1/3 Inch optical format