The University of Adelaide CENTRE FOR AUTOMOTIVE SAFETY RESEARCH

home   /   centre for automotive safety research   /   Publications   /   List   /   Details

Publication Details

TitleUsing cyclist detection software to count cyclists in video and measure their speed
AuthorsWoolley JE, Szpak ZL, Ponte G, Searson DJ
Year2014
TypeConference Paper
AbstractThe Australian Centre for Visual Technologies recently developed software for post-processing video footage that is capable of detecting cyclists and assessing their level of conspicuity. In partnership with the Centre for Automotive Safety Research (CASR), a pilot study has been conducted to validate the ability of the software to count cyclists and determine their travel speed. Commonly available GoPro video cameras were used to film cyclists from various angles in a controlled setting. The speeds of the cyclists were recorded using GPS data loggers as well as being measured using a laser speed gun. The cyclist counts and speed measurements determined from the video detection software correlated well with the true count and the speeds measured with the laser gun and GPS. Following the successful pilot study, three real world site investigations were conducted. At each site, two GoPro cameras were used to film cyclists travelling over specialised bicycle detection counters installed by the local transport authority. This enabled a comparison between the counts provided by the video detection software and the counts provided by the closed induction loop counters. The results indicated good correlation and demonstrated the feasibility of using readily obtainable video footage to obtain objective bicycle data.

Many uses for this technology can be envisaged in future work. In the simplest case, video detection software may provide a means of inexpensive cyclist counts that does not require a complicated installation; in fact even existing traffic management video footage may be used. Further, if the geometry of a site is known, other applications for this software include measuring cyclist speeds and lane positioning. Research is currently underway to develop capabilities for identifying near crash incidents on roads including cyclist-vehicle, and cyclist-cyclist as well as cyclist-pedestrian interactions on shared paths.

SponsorMotor Accident Commission
Conference NameAdelaide Velo-city Global Conference 2014
Conference LocationAdelaide, South Australia
Conference Date27-30 May 2014

Reference
Woolley JE, Szpak ZL, Ponte G, Searson DJ (2014). Using cyclist detection software to count cyclists in video and measure their speed. Adelaide Velo-city Global Conference 2014, Adelaide, South Australia, 27-30 May 2014.