Tomasz Gutowski has used yesterday's timelapse tutorial to collect and analyse data on pedestrian movement. Using the timestamps from the imagery and vector based positioning to create a density grid the results are remarkably impressive.
Toms YouTube movie below provides full details:
The possibilities are notable and Tom should be congratulated on the methodology...
I know that there are some commercial solutions that "trace/identify" the heads and retrieve the density of pedestrians automatically. Do you know if there is any free / academic solution or, alternatively, image recognition method that can be adapted for this? It seems that he did manually...
It does seem to be manually done - but if you can use the timestamp images and autodetect then it would be a quick and easy way to implement density grids.
We are going to look into it and I'll post and thoughts soon as i can...
Hey Andy,
ReplyDeleteI know that there are some commercial solutions that "trace/identify" the heads and retrieve the density of pedestrians automatically. Do you know if there is any free / academic solution or, alternatively, image recognition method that can be adapted for this? It seems that he did manually...
It does seem to be manually done - but if you can use the timestamp images and autodetect then it would be a quick and easy way to implement density grids.
ReplyDeleteWe are going to look into it and I'll post and thoughts soon as i can...
Andy