Technology Transfer

Vision-Based Pedestrian Detection System
  • Year:2015
  • Project:DOIT Program
  • Category:Imaged assist technology
  • Technique Features:

    The vision-based pedestrian detection system uses monocular camera to capture image information in front of vehicle. This system employs pedestrian extraction feature method to transform images to be the descriptor; meanwhile, it can automatically detect pedestrians and send alarms when classifying and facing pedestrians.

  • Specification:

    1.Frame per second:10 frame/s以上
    2.Display resolution:720*480
    3.Embedded hardware:FPGA;Operating frequency:200 ~ 800 MHz
    4.Interface:CAN BUS (Steering sensor、Gear position),NTSC (Camera)
    5.Pedestrian detection function: Behind vehicle between 5 to 15 m
    6.Operation temperature: -40℃ to 85℃
    7.Operation current: CCD(12V,120mA)

  • Applications:

    Advanced Driver Assistance System (ADAS)

  • Benefit:

    The vision-based pedestrian detection can effectively decrease the risks of inappropriate driving. With this system, drivers may obtain pedestrian information in front of vehicles, which are key point on driving. This field would be treated as a major profit for electrical companies that stress on vehicle interventions.

  • Contact:

    rdservice@artc.org.tw