deep learning vision processor with data security and open ecological chain
5M professional security video quality and 1.5TOPS computing power@INT8
Peak Performance
Video Decoding
Professional Security Image
Processor
Self-intellectual property tensor processor
Professional security image
Complete Deep learning Toolchain
Safety processor architecture
High integrated high performance platform
Smart camera, wisdom analysis box, face recognition attendance machine, face recognition access control, smart traffic testing provides intelligent visual front end for the security industry.
Smart garbage sorting, smart city, smart campus, wisdom apartment, etc., providing visual calculation target index tracking.
Obstacles perception, reverse image, parking unit detection, driver behavior detection warning, rear mounting soft hard integration scheme, etc. For security assist driving to provide immediate power guaranteed.
Intelligent unmanned retail, product test identification, no blush face payment, etc. Provide key end-side computing links for intelligent retail.
Educate robots, inspected robots, welcome robots, entertainment robots, etc., providing neural network machine visual capabilities for all industrial automation terminals.
Intelligent industrial production line defect testing, fault testing warning, smart product sorting, etc., providing safe assistance for intelligent industries, providing production efficiency.
Product Features
CV1835 professional intelligent Ip Camera SoC
Processor core
Dual-core RISC-V
Self-developed TPU
Support mainstream neural network frameworks: Caffe, TensorFlow, TensorFlow Lite, Pytorch, ONNX and MXNet
Video Encoding and Decoding
5M
H.265 Main Profile
H.264 Baseline/Main/High Profile
JPEG
Support Issue (ROI) Coding
Image and Video Processing
Image and video 90-, 180- and 270-degree rotation, Mirror, Flip
3A (AE / AWB / AF) Algorithm
Regional Adaptive Defog
3D noise reduction, detail enhancement and sharping
Sensor belt wide diversion and 2 frame wide dynamics
Fisheyard
Audio codec performance
Software audio codec protocol(G.711、G.726、ADPCM)
Support audio 3A (AEC, ANR, AGC) function
Video interface
Input:
Support two-way inputs
Support MIPI, Sub-LVDS, HiSPi interface
Support CMOS sensor
Output:
Support 1 output interface
Support MIPI-DSI output interface
Audio port
Integrated Audio Codec, supporting voice input and output
Support I2S/PCM/TDM interface, docker external Audio codec and provide MCLK clock output
Safety
Support safe start and upgrade
Hardware plus decryption
Peripheral interface
Integrated GPIO, ADC, I2C, SPI, UART, PWM, SDIO 3.0, USB 2.0 Host/Device
Integrated Key scan, Wiegand
Integrated MAC PHY, supports RMII mode; 10/100 Mbps full duplex or half duplex mode, supports RGMII mode; support 10/100/1000Mbps
Memory interface
Support DDR3/DDR4/LPDDR4
SPI Nor Flash interface
SPI Nand Flash interface
Support eMMC4.5 interface
Physical specifications
Support multi-level power saving mode
Package form 13.5mm x 13.5mm,BAG package size, pin spacing: 0.65 mm pitch