Join or Sign in

Register for your free asmag.com membership or if you are already a member,
sign in using your preferred method below.

To check your latest product inquiries, manage newsletter preference, update personal / company profile, or download member-exclusive reports, log in to your account now!
Login asmag.comMember Registration
https://www.asmag.com/showpost/32521.aspx
INSIGHTS

Rockchip launches new chips for edge and back-end security solutions

Rockchip launches new chips for edge and back-end security solutions
Rockchip has introduces its new security solutions including: inteligent vision processors RV1126/RV1109, focusing on front-end products and RK3568/RK3588 solutions focusing on back-end NVR+XVR products.
Semiconductor designer Rockchip introduces new security solutions including: intelligent vision processors RV1126/RV1109 for edge devices such as industrial PCs, smart door bells and smart gates and access control. The second set of Chips RK3568/RK3588 are designed for back-end NVR+XVR products.
 
Rockchip is market-oriented, providing customers with multi-level, multi-platform and professional solutions. It has accumulated rich experience in image visual processing, video coding and decoding, SOC chip design and system development. The new generation of products tackles the pain points of the security market and the requirements of different applications. Rockchip helps optimize technology performance and enhance the market competitiveness of security products. 

Technical Features of RV1126/RV1109

The newly launched front-end security product solutions, RV1126 and RV1109, both adopt 14nm process technology and have the NPU and ISP built-in. The hardware has a pin 2 pin structure and supports easy product upgrades.
 
Key specs of RV1126:
  • Quad-core Arm Cortex-A7 CPU
  • 2.0T NPU
  • 14M ISP with 3F HDR
  • 4K H.264/H.265 30fps video encoder and decoder
 
Key specs of RV1109:
  • Dual-core Arm Cortex-A7 CPU
  • 1.2T NPU
  • 5M ISP with 3F HDR
  • 5M H.264/H.265 30fps video encoder and decoder

The RV1126 & RV1109 provide six main technical features:
 
I. See clearly
 
RV1109 and RV1126 provide temporal noise reduction and multi-frames HDR technology for camera image processing, capturing more detailed images at high dynamic scenes and clearer color images at dim light scenes.

II. Hear clearly
RV1126 and RV1109 have built-in HD AEC algorithms, support microphone array, enhanced voice acquisition and extended pickup distance.
 
III. Smaller memory space
Both chips support low bit rate, low delay coding and perceptual coding, reducing needed storage space and improving user experience.

IV. Higher AI computing performance
Rockchip provides a complete tool RKNN-Toolkit that supports heterogeneous computing and mixed-precision quantization, and flexibly supports various network model architectures. Rock-X SDK, an AI algorithm library, is also provided.
 
The inference power consumption of the same model on the NPU is less than the 10% required on the GPU.
 
AI models generated by Tensorflow, PyTorch, Caffe, MxNet, DarkNet, ONNX and other frameworks can be directly converted and deployed.

V. Customized cameras
RV1126 and RV1109 support fast boot, dual OS and open source SDK. Customers can make in-depth customization to enhance the competitive advantage of products.

VI. Support mainstream encryption and decryption engines
Both chips support TrustZone technology, several encryption engines including that of NCA, and secure boot to ensure the security and integrity of data.
 
Moreover, RV1126 and RV1109 have ample interfaces, including Gigabit Ethernet interface, supporting TSO network acceleration; USB 2.0 OTG and USB 2.0 host, two SDIO 3.0 interfaces, 8-channel I2S with TDM/PDM and 2-channel I2S to fulfill the diversified needs of a variety of security applications.
 
The front-end security products, target three main categories: IPC, battery IPC/Smart visual doorbell and smart gate/access control.
 
For IPC products, the RV1126 and RV1109 provide three technical advantages:
  1. Multi-frames HDR technology can ensure detail presentation in HDR scenarios, and acquire high-quality color images at dim light scene thanks to temporal noise reduction technology
  2. Intelligent coding technology is more efficient in achieving low bit rate coding and perceptual coding
  3. Built-in NPU, mature AI tools and rich AI ecological algorithms enable stronger computing power and greatly expand the possible applications
For battery-powered IPC/smart visual doorbell products, customers' core demands are longer battery life, quick snapshots and HD voice call, etc. RV1126 and RV1109 provide the following technical advantages:
  1. Support fast boot, and 150ms cold boot for quick snapshots
  2. Extended battery life (dual OS, low-power architecture, 14nm process, LP3/LP4 DDR)
  3. Built-in NPU to filter invalid alarms
For smart gate/access control products, RV1126 and RV1109 provide faster and safer facial recognition:
  1. Support dual MIPI sensor input, achieving more sensitive perception, and
    support Face Liveness Detection, preventing picture or video validation
  2. ISP optimization for back-light and front-light scenarios ensures accurate recognition in poor light environments
  3. Diverse AI algorithms of RV1126&RV11109 extend product applications

Technical Features of NVR+XVR product solutions RK3568 and RK3588

Rockchip has also launched two mid-end and high-end solutions - RK3568 and RK3588 for back end NVR and XVR
 
Key specs of RK3568:
  • Quad-core Cortex-A55
  • Mali-G52 GPU
  • 0.8T NPU
 
RK3568 has friendly software interface suite design: the API is packaged by modules (such as AI/AO/VO/VDEC/VENC) to help customers develop application software rapidly.
 
 
There are eight main technical features of RK3568:
  1. Supports 10-channel 1080P 30FPS video decoding
  2. Supports TSO/UFO hardware acceleration, enhances the access capability of webcams
  3. Built-in 0.8T NPU hardware accelerator to support face comparison, face search, alarm, human-vehicle classification. It supports one-click conversion and Caffe/TensorFlow/TF-Lite/ONNX/PyTorch/Keras/Darknet mainstream architecture models.
  4. Multi-channel display interfaces: supporting dual/three-screen display, HDMI2.0b, 4K image quality, and eDP/MIPI DSI/LVDS/RGB888/BT.1120/BT.656;
  5. Powerful post-processing capability: It supports 3D-LUT/GAMMA/
    CSC/BCSH/DITHER/CABC, and image post-processing, de-noising and color enhancement.
  6. Strong decoding capability: Supports 4K H.264/H.265/VP9 HD decoding in multiple formats, and ROI coding
  7. A strong compatibility of DDR particles: Supports LP4/LP4x/LP3
    /DDR4/DDR3, and DDR3 and DDR4 ECC for high reliability requirements of industrial control scenarios. With bitwidth of 32 bits, maximum frequency of 1600MHZ, and design of ample bandwidth, it is able to meet the requirements of intelligent NVR security scenarios
  8. Rich peripheral interfaces: It supports 3xSATA3.0dual Gigabit network port, TSO technology, dual PCIe, which can meet the expansion requirements of 4G/5G/WIFI6, multi-network ports and NPU.
 
Key specs of RK3588:
  • 8nm process
  • Quad-core Cortex-A76+Quad-core Cortex-A55
  • Quad-core GPU
  • 6T NPU

RK3588 is a high-end solution for NVR+XVR products. It offers advantages compared to current products on the market that possess low video decoding capacity, insufficient NPU computing power, less display interfaces and less high-speed interfaces. RK3588 has the following FOUR advantages:
  1. It supports 32-channel 1080P video decoding
  2. It supports a 4-screen display
  3. It has multi-channel high-speed interfaces: supporting multi-channel PCIe3.0/PCIe2.0,SATA3.0, USB3.1 and Gigabit Ethernet
  4. Powerful computing: 6Tops NPU, Octa-core CPU, Quad-core GPU

Rockchip Efficient Developer Tool – RKNN Toolkit

In addition to the above new powerful solutions, Rockchip has also introduced a more efficient and accurate developer tool - RKNN Toolkit.
 
The functions of RKNN-Toolkit include model conversion, performance and memory evaluation, simulation, accuracy analysis, model visualization, model pre-compilation, model segmentation, batch inferencing, mixed-precision quantization, model optimization, etc. Through GUI, RKNN-Toolkit makes developers' work easier.
 
RKNN Toolkit supports two quantization mechanisms: post-training quantification and quantification-aware training.  RKNN Toolkit, coupled with a flexible multimedia development interface, can increase the flexibility of developers' self-defined applications, and help partners expand more AI functions and applications.

For more information, please visit the company website here.
 
Subscribe to Newsletter
Stay updated with the latest trends and technologies in physical security

Share to: