9. Large Multi-Modal Models (LMMMs) to Incorporate Images, Art & Video -[E3] Naik, Nikhil, Scott Duke Kominers, Ramesh Raskar, Edward L. Glaeser, César A. Hidalgo

[lkcao]

Post questions here for this week's exemplary readings:

3. Naik, Nikhil, Scott Duke Kominers, Ramesh Raskar, Edward L. Glaeser, César A. Hidalgo. 2017. “Computer vision uncovers predictors of physical urban change.” PNAS 114(29):7571–7576.

[floriatea]

Beyond economic and demographic indicators, how might the aesthetic elements of urban environments, as captured through computer vision, serve as predictors of social and economic mobility within communities?

[michplunkett]

A lot of research papers that we've read so far in class, this piece included, have used advanced computational methods to assert theories that have long felt intuitively and/or anecdotally true. Is this sort of research trend the norm for younger academic fields, or is this unique to ML?

[Caojie2001]

I appreciate the authors' work. However, similar to @michplunkett's point, the result of the paper seems to be a little bit boring for me, as it only provides empirical evidence for an argument that has already been well-known. I wonder whether any development directions would make the conclusions of this study more valuable academically or practically.

[alejandrosarria0296]

Recent understandings of how neighborhoods get gentrified tell us that the property ang zoning laws tend to be the first step towards rising rent prices. Could the method applied by the authors be used in conjuction with legal data to cdetect an "image" of gentrification?

[Twilight233333]

The author's research is very interesting and I would like to know how the author captures the changes in business and whether commercial buildings are isolated in the empirical research. How to eliminate the impact of large-scale road construction. How to control the time impact of photo capture.

[yueqil2]

This article is easy to understand as a whole, and the data and methods part does not seem to be as full of mathematical principles as the articles I read before, which is quite different from my expectations. What algorithmic breakthrough did this research achieve? Or is it just an application?

[HamsterradYC]

With rapid advancements in imaging technology and artificial intelligence, what are the future research directions for utilizing these tools in urban studies? How might emerging technologies like deep learning further refine the accuracy of urban change detection?

[Carolineyx]

I really like how they use the photo to predict community change, I wonder besides recognizing objects in the photo is there anyway to extract the emotions in the photo, or ambiances of the geolocation?

[Brian-W00]

How might the presence of green spaces and parks within urban neighbourhoods influence those areas' physical improvement and safety perceptions?

[Dededon]

This is an interesting paper to combine satellite images and neighborhood effects. I'm curious about whether this research could be extended by counterfactuals: could neighborhood status deteriorate after being a decent-and-gentrified neighborhood?

[JessicaCaishanghai]

This paper is very interesting. However, I just wonder how the grass building and sky can be read of signs of safety? I think it really depends. Maybe an area can be very green and unsafe at the same time, and the sky is mostly related to the local weather.

[icarlous]

What I am thinking about is connecting this urban data with digital trace gathered from online. How could the cyber space interact with the actual one?