PHOTOGRAPHIC SURVEYING:
HONG KONG
WALKABILITY
PROJECT
2019 - 2021
When you are healthy, in shape and alone, walking the streets of Hong Kong doesn’t feel that hard. Even the narrow sloppy sidewalks in Central may be a breeze for you. But then, the clock strikes twelve, and it's lunch time for everyone. The streets that you found easy to zip through are now full of people, bustling with life (and random trolleys), and you are suddenly a parent with a stroller and a wailing child in your arms.
Does it still feel easy to walk in Hong Kong -- in other words -- are the streets of Hong Kong still walkable? Online maps may readily provide geographic information, but they do not really inform users about the walkability of that area.
The main assignment of CCGL9061 from 2019 - 2021is a large-scaled project that focuses on the walkability of Hong Kong, of which all students take part to collect data. There are several steps for this large-scale project:
1
GROUP DIVISION &
SELECTING AREAS
The students are divided into groups of 5 or 6. They then select one subway station and share the task of covering all the streets and alleys in this area.
2
ON THE STREETS:
COLLECTING DATA
The students then pick up their most comfortable shoes and their smartphones, and burn some calories while taking geo-tagged pictures every 25 meters and on both sides of the streets in the areas they have chosen.
3
WALKABILITY JUDGMENT TASK
(TRY THE DEMO!)
Then, with an online tool created for the purpose of the project, students assess the walkability of the locations seen on the pictures taken. How exactly do they do that? They simply look at two photos presented side by side on screen, and pick the one where it seems easier to walk. Repeat this small judgment task a few hundred times… Don’t worry, it doesn’t take that long!
Try the online tool demo above!
4
ANALYZING JUDGMENT TASK RESULTS
Then let our teachers compute the students' judgments into walkability "elo" scores (the way tennis players get ranked by their wins and defeats). Then we try to work some deep learning magic to train the computer to automatically assess walkability for any image of the streets.
5
GROUP VIDEO
REPORT
Students then create a short group video reporting on the area surveyed and the challenges met. The more relevant information it contains and the more entertaining it is, the better! And once again, don’t worry, some tutorials are here to guide their efforts.
WATCH OUR VIDEO REPORTS
WHAT COMES NEXT?
Interim results & progress updates:
Apart from developing students’ digital literacy with video creation, another important objective of the project is to lay down the foundations to constructing digital maps that display the walkability of different areas of Hong Kong with coloured indicators. One can then assess how different areas differ from each other, and also how quickly things may change from one street to the other.
We are currently training deep neural networks to try to predict walkability from any picture of a street. The task is challenging and we only have little data, and we try different state-of-the-art architectures to get the best results possible. Read through the report below to learn more.
