For those of you who don't know, the current state of the art in determining the number of cyclists and pedestrians utilizing the streets is to put a volunteer on a street corner, and count them. For cars, we have those nice rubber bands you see placed across the street. Sad, right?
I am hoping to improve on this situation by using the abundance of high quality satellite imagery available to count pedestrians, bicycles, and cars in a given image (just as a jumping off point) in order to determine modeshare.
Putting the various issues that will eventually arise when analyzing this data, I was wondering if any of you might be able to help me find libraries that would help me perform this sort of analysis. Any sort of jumping off point, existing research, similar projects, really anything you can provide would be helpful.
I am simply a web developer interested in alternative transportation and public policy who believes there must be a better way (cell phone location data would be nice, and better, but it is hard to get). The friendlier the language is with the web (python, javascript - Atwood's law anyone?, php, etc...), the better as what I would really like to build is a web service that allows policy makers to perform custom searches, and do their own analysis online.
I am definitely willing to learn whatever I need to to optimize my results, but I'm starting from pretty much nothing when it comes to pattern recognition so please keep in mind I may not understand the vocabulary that goes along with it.
If it means anything to anyone, I have access to ArcGIS Server if needed. I am hoping to submit this project for my Intro to Geospatial Technologies course, due late February. I plan to use imagery from my university, RIT, as a starting point as our Imaging Science department happens to like to take lots of pictures of campus from space.
[link][2 comments]