본문바로가기
  • HOME
  • ADAS Processor
  • APACHE5

Deep Learning based Edge Processor
ISO26262 (ASIL-B) Compliant

APACHE5 (NVS2900) is the vision-based Edge Processor with NN accelerator(NPU) for ADAS and AD application. It builds on the success of APACHE4, featuring upgraded Quadcore A53 CPUs and more advanced imaging engines as well as significant NN accelerator for camera upto 5.7M resolution.The integrated NN accelerator is capable of delivering 1.6 TOPS performance at sustained efficiencies up to 95% with very low power consumption. The highly integrated set of capabilities on the APACHE5 will enable Tier1s and OEMs to address some of the most demanding challenges of autonomous automotive applications such as Rear/Front AEB, DSM/OSM and FSD.

• NPU : Neural Network Processor Unit
• TOPs : Trillion Operation Per Second
• CNN : Convolution Neural Network
• AEB : Autonomous Emergency Braking
• DSM : Driver Status Monitoring
• OSM : Occupant Status Monitoring
• FSD : Free Space Detection
• ADAS : Advanced Driver Assistant System
• AD : Autonomous Driving

  • Resolution

    5.7M(3072x1876) @30fps
    3.8M(2650x1440) @45fps

  • Core

    ARM Cortex-A53 (Quad Core)@1GHz (3,200DMIPS)

  • Embedded Vision Processor

    AIMOTIVE aiWare CNN Processor(1.6TOPS)

  • Video In

    4-lane MIPI CSI-2 Tx for interfacing with ADAS ECU or other Video Devices

  • Video Out

    MIPI-CSI2 4-Lane/Parallel

  • Function

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

  • Size

    9mmⅹ9mm, 0.5mm pitch 289-BGA

SCROLL DOWN
Edge Processor
Based on Deep Learning Algorithm

Deep learning is the most popular advanced technology for AI artificial intelligence.
It is known as an in-depth learning method using a multi-layered CNN that mimics human neural networks to perform machine learning.
APACHE5 embeds a dedicated processor NPU that enables this deep learning technology computation.
The built-in NPU has a performance of 1.6 TOPS and, above all, its performance is 98%(the utilization ratio) and its performance as an NPU is one of the best in the industry.

Perfect computational NPU
for Automotive CNN inferences

Machine learning is a technology that learns important patterns and rules from large amounts of data and makes decisions and predictions based on them.
This technology iterates training and inference and goes to the ultimate solution.
The computing power for the inference model is provided as a powerful NPU by the edge of APACHE5.

Apache5 can be used in
wide range of camera application for ADAS
including

- Driver Monitoring System(DMS)
- Smart RearView(SRV)
- Autonomous Emergency Braking(AEB)
- Lane Departure Warning System(LDWS)
- Lane Keeping Assist System(LKAS)

Exclusive development tool
to implement deep learning technology

aiWare Studio
: Performance Measuring and Profiling of developed CNN Network
: Visualize neural network structure and individual layer details
: Create binary files for NPU

Runtime SDK
: Supports various frameworks
: Enables development of various applications

SITE MAP