Robust Object Detection System

I&T Solution Robust Object Detection System
(REF : S-0965)
Trial Project
Solution Feature
  • Video of tramway track at 5 to 22 meters in front of a tram is recorded by a self stabilising camera
  • Video footages are sent, through wired connection, to an industrial computer installed on the tram to be converted into a sequence of images at a certain framerate
  • An image is analysed by a convolutional neural network on the computer to detect the presence of objects of varying sizes on tramway tracks
  • Upon successful detection of obstacle on tramway track, an image of the obstacle will be displayed on a visual panel before the tram driver and an alarm is sounded
  • Tram driver needs to press a button on visual panel to stop the alarm, and to press the brake to stop the tram for safety
Trial Application and Expected Outcome
  • Creating object detection servers on two edge computers: for each server, build an object detection program by learning a convolutional deep neural network from image data for tramway tracks with obstacles
  • Installing and testing object detection systems (ODS) on two trams in the route between Shau Kei Wan and Happy Valley: set up cameras, panels, and computers (one month); deploy all ODS's (four months); check for errors from the logs recorded in the computers and fix any errors
  • Training and real-life testing: train tram drivers to respond to alert signals produced by the ODS's; let tram drivers use ODS's for ten months; collect survey results from tram drivers on their experience of using the ODS's
  • Evaluating real-life trials and updating object detection system: estimate the range of drivers' reaction time from the survey results; collect relevant diagnostic data from survey results; update the ODS's using diagnostic data, e.g. re-compute cameras' viewing distance and adjust their line of sight (3 months)
  • Expected Outcomes: two trams in the route connecting Shau Kei Wan and Happy Valley in Hong Kong Island will successfully be installed with an ODS; two tram drivers will have been trained to use an ODS to assist in their driving
Info on I&T Solution Provider
Address:Rm 2, 7/F., Block B, Mai Hing Industrial Building, 16-18 Hing Yip Street, Kwun Tong, Kowloon, Hong Kong
Contact Person:ALEX MAR
Tel: +852-9512-0604

For details of the above I&T solution, please contact the I&T solution provider.