After reading a number of research and guidance on pedestrian counting. I am familiar with the basic requirements and conditions. Yet, I also noticed that some leading scholars using cameras to audit pedestrian while simply walk the full length of each street segment, this undoubtably lower labor cost and less time-consuming. First, I think the duration of time might not be sufficient, and variation might be strong. Yet, as for measuring pedestrian concentration and distribution, this method might be able to help since it is much faster under the condition of peak time duration and normal weekday. I also conducted a tentative counting based on this method. the variation is not that strong as I expected. So, do you think it is plausible to audit the number of pedestrian walking pass through you by walking the whole length of a street segment?
Dear Chris,
in Poland street cameras are commonly used to count pedestrian's flow. The flow is analyzed by manual counting by the observer watching the on-line view from the camera between 6 am to 9 pm or even longer if the observed place is interesting enough to lure people also at night.
Nowadays more and more researchers are using computer programs which are able to recognize faces to count the flow of pedestrians because it is much faster.
Remeber that the camera/s must be set permanently and show the same area throughout the whole survey. It is good to conduct study for a longer period of time - several different days with different weather conditions. Bad weather (not extremely bad) can lower the flow by even one third.
Good luck!
Dear Hubert,
Thank you so much for the comments. Yes, commonly we all tie the camera permanently to observe pedestrian flow. So, it is really baffling to see scholars using "walk the length" method to measure pedestrian distribution. It is really nice for you to clear that up.
Thank you. Wish you all the best.
Best,
Chris
Hi, maybe I can add something. There is a similar data collection technique, where pedestrians counts can be done from a moving vehicle using video capture every 100 m or so. The idea is to count the peds in the visible road section ahead. In slow and heavily pedestrianised traffic conditions such pedestrian walking and crossing count represents a sample in a known time period (eg 10 sec). Such can be calibrated to peak hour flows against a known 24 h distribution. This is a rough approximation of pedestrian activity only. I'd like to know if someone has actually used it successfully. How well does it calibrate vs. manual peak flow counts? I'd expect high variability.
- Badges
- Science topic