본문바로가기
  • HOME
  • ADAS Processor
  • APACHE4

Computer Vision based Edge Processor
ISO26262 (ASIL-B) Compliant

APACHE4 (NVS2800) is the vision-based Edge Processor for ADAS and AD application. It leads to achieve the real-time operation by hardwired its sensing algorithms as Computer Vision as well as Deep learning operation with DSP(CEVA-XM4). In addition, as a complete edge processor, it reduces the load on the central ECU of ADAS/AD system and enables to select affordable ECU line-up. Advanced image processing technology powered by Nextchip is integrated to provide a dramatically improved recognition rate.

• ADAS : Advanced Driver Assistant System
• AD : Autonomous Driving
• DSP : Digital Signal Processor
• ECU : Electronic Control Unit

  • Resolution

    5.7M(2650x1920) @30fps

  • Core

    ARM Cortex-R5F (Dual Core. 2,338DMIPS)

  • Video In

    MIPI-CSI2 4-Lane/Parallel

  • Video Out

    MIPI-CSI2 4-Lane/Parallel

  • Function

    HDR, LFM, 2D/3D-NR, LDC, PGL

  • Recognition Engine

    CV base Multi-class detector (Vehicle, Pedestrian, Lane and Moving Vector)

  • Size

    9mm x 9mm (0.5p)

SCROLL DOWN
Edge processors
Based on computer vision algorithms

APACHE4, a processor that detects vehicles, pedestrians, lanes, etc., is a device suitable for smart sensing cameras for ADAS system.
It is designed based on computer vision algorithms and realizes low power and high efficiency sensing technology. Problems that conventional smart cameras have can be solved even if in tricky corner cases (e.g., pedestrians wearing black suit in black background, pedestrians wearing white suite in white background, etc.) and in the extremely low light condition.

World Top Class ISP technology
onto APACHE4

The pre-condition for good object detection is clear image. APACHE4 is the first generation edge processor that combines Nextchip's several years' worth of image processing know-how with sensing technology for the first time.
Built-in image processor up to 5.7 megapixels delivers the highest image quality and sensing performance as an edge processors.

Securing Real-time processing
by designing ASIC of sensing algorithm

APACHE4 enables real-time detecting of various objects at the same time with higher resolution using hardwired Computer Vision (CV) algorithm.
In other words, Edge does the functions that were performed in the existing central ECU and it enables high performance real-time computation processing, thereby reducing the load on the ECU and making it competitive in cost way.
This Edge, decentralization architecture of smart functionality, allows to apply ADAS functionality in various automotive line-ups even affordable cars.

In-house developed program

- NextADAS
- APACHE4 SDK
The most labor-intensive part of validating ADAS solution is a data collection as well as multiple learnings.
NextADAS is even used for Annotation, a very initial step of data collection.

SITE MAP