Video-Assisted Bicycle Data
What if we could quickly and inexpensively gather detailed bicycle data – including the many customizable details that are important to us – quickly and inexpensively, thanks to today's easy availability of video, and using special software to quickly analyze it? Compressing long hours of observation into the minimum, while being able to gather even more information than ever before!
Good bicycle data is essential to advancing the cause of bicycling, but is too often not available, and is typically a huge job to obtain. Funding and other policy decisions are frequently made based on the limited bicycle data available. Too often, bicycling is neglected, misunderstood, or excluded from consideration as a result. We need good data!
In the bicycle world we care about much more than just how many are bicycling; we care about everything – the gender ratio, age range, behavior, style, and practical information such as whether items are being carried and how. There is so much that makes bicycling unique, to help us answer key questions to understand, improve and provide for bicycling. The typology of bicyclists is diverse, and the information sought will be quite different from one researcher to another.
The purpose of the Atlantis Viewfinder project is to make new software analysis tools widely available as open source software, to dramatically increase the quality and quantity of data available. Manual recording of bicycling is too error-prone and labor intensive to carry out on any scale. New technology exists to count individual bicycles, but is also limited and expensive.
Video detection allows very long periods of automatic data collection to be processed in a much shorter time. Much as a file can be compressed to a much smaller size, with video detection we can jump past the empty space in the video so that the processing time of even a very long video is shortened considerably. Ten hours can be done in ten minutes, if few bicycles have passed. Even better, we can avoid the pollution and fatigue of standing on street corners, and just one person can set up cameras to start simultaneously, enabling one to scan many sites at once (important for uniformity and comparability of data).
The software scans the video, identifying moving objects that appear to be bicycles and automatically recording time of day, direction of travel, and where possible, speed, plus generate an initial bicycle count. Need more?
We can then jump instantly from bicycle to bicycle, skipping over empty spaces, entering additional data we wish to capture. We can even collect data about other types of travelers, such as scooters, skateboards, baby strollers, and pedestrians.
Want to go deeper into the bicycle experience? You customize the data entry, for rapidly collecting further details of your choosing!
Want to record all the cargo bikes? Kids on bikes? Kids in bike seats? Bike seats with shopping bags instead of kids? People wearing helmets? People wearing “no helmets” buttons”? Cool cats with both handlebar moustaches and moustache handlebars? Ice cream cones by group size on summer days? Smiles and frowns, or any signals? Subjectively assess how relaxed the cyclist appears, or their experience level? All of the above? Just say so!
This is the data that a computer cannot automatically determine, but the possibilities for an interested human being are quite limitless. In the field, it isn't possible to write a long list of observations by hand, particularly if bicyclists are coming quickly. Errors are inevitable, and missed data assured. Video lets us take more time when we need it, and to jump ahead when we don't. So enter all the data items you're interested in!
Atlantis Viewfinder is in development as a project of Escher City Associates, led by Jason Meggs, a computer scientist, bicycle researcher and longtime bicycle booster, most recently working on bicycle data collection techniques with the BICY project, an EU-funded study intended to increase bicycling in Central Europe and beyond. Very familiar with the limits of inadequate data to bicycle policymaking, Jason has spent many hours on traffic-clogged corners tabulating passing bicyclists, as well as many more hours with bicycle-related video; to combine the two seemed natural and a boon to the bicycle effort. Assistance to the development effort is most welcome.