Recommend AI box with built-in video analysis algorithms

     AI platform

  • Features
  • Specification
  • Solution
  • Picture

typical models are 8/16/32/64/128 channels, receive RTSP stream as input, alarm/alert report in smarteye WEB client.

smart video analysis algorithm supported

face, human, vehicles, fire&smoke, leave-post, high-voltage-warning-sign, fire-extinguisher, smoking, zone-forbidden, fence-climb, helmet/hardhat recognition, safety-vest, phone-call, sleep-on-post, working-clothes, falling, congregation, accumulation, Camera-obstruction, motion-detect, fire-exit-occupation etc.

SPEC of typical 16channel model

CPU

Quad core Cortex A53 NEON accelleration ,  FPU built-in

NPU

30TOPs@INT4  with  CNNRCNN  etc

RAM

8GB DDR4

ROM

8GB EMMC 5.1

DSP

800MHZ VDSP

system

Linux 

Video encoder

H.264/H.265  with I/P/B frame, 3840*2160@60fps  support QP CU map/IPCM CU/skip CU , max bitrate 200Mbps 1080Px8

Video decoder

H.264/H.265 , max decoder 300Mbps4K@60fps,  support I/P/B, max 16 reference frame, max 1080Px16

Security

Safe mode boot supported

encryption

HW  AES/DES/3DES SHA-1 / SHA-224 / SHA-256

Video out

 HDMI 2.0b output

audio

MICx1 external  Audio Codec supported

network

Gigabit ethernet RJ45 x 2

 

smart video analysis algorithm supported

face, human, vehicles, fire&smoke, leave-post, high-voltage-warning-sign, fire-extinguisher, smoking, zone-forbidden, fence-climb, helmet/hardhat recognition, safety-vest, phone-call, sleep-on-post, working-clothes, falling, congregation, accumulation, Camera-obstruction, motion-detect, fire-exit-occupation etc.