Matt

cs phd candidate @cornell_tech
urban computing & normativity

[Matt @mattwfranchi]

We’re releasing a new research dataset of 2.4 million street-level dashcam images, spanning all five boroughs of New York City from March-November 2020.

https://t.co/eYonxMZ2vf

[Emma Pierson @2plus2make5]

New paper in Nature Communications! We show you can map out urban flooding by using vision-language models to detect floods in large-scale street scene datasets.

In New York City, our method identifies flooded neighborhoods, home to 100k people, that current methods miss. 1/

[Matt @mattwfranchi]

@jialiiiz Sorry, we don’t know each other but your work came up on feed; seems super cool (!), and also very Hofstadter’s « analogy as the core of cognition »

https://t.co/5iPlVepiLB

Very happy to see work like this coming out.

[Maria Teresa Parreira @mteresaparreira]

Are you an expert in robot navigation, human-robot interaction, urban planning or accessibility? STOP SCROLLING!

https://t.co/Tv1E0svUZh
(est. time < 20 min)

[Megan Richards @megan_richards_]

Very excited that this is now out! A broad group across FAIR and academia (> 40 people across 10 institutions) came together to help democratize knowledge about VLM training, responsible evaluation, and promising research directions (including work in video!).

[Frank Bu @FrankBu0616]

#CHI2024 Our paper Portobello: Extending Driving Simulation from the Lab to the Road is now available online. The Portobello infrastructure system helps researchers port in-lab driving simulations to on-road environments.
https://t.co/1cfXNA3bUe

[Matt @mattwfranchi]

Use this link to submit a data access request; we will provide access to non-commercial researchers who consent to the Dataset's Terms of Use.

https://t.co/eYonxMZ2vf

[Matt @mattwfranchi]

We’re releasing a new research dataset of 2.4 million street-level dashcam images, spanning all five boroughs of New York City from March-November 2020.

https://t.co/eYonxMZ2vf

[Matt @mattwfranchi]

This dataset is broadly useful for research in computer vision, urban studies, and the social sciences (as well as replication of our original study). It has high temporal granularity, with tens of thousands of images per day.

[Matt @mattwfranchi]

Excited to share our new preprint, Fingerprinting NYC's Scaffolding Problem with Longitudinal Dashcam Data (https://t.co/Xyi8OyfmkU) [in review KDD'24].

Read a synposis about the work in @TheEconomist! Excited to share more post-peer review.

[The Economist @TheEconomist]

New Yorkers have grown so sick of ugly scaffolding that last year the mayor launched a campaign to clear it from the streets.

Now computer scientists have made a hit list of possible targets using AI https://t.co/hCK4s4SfIZ 👇

[patrick. @imPatrickT]

not a phone in sight

[jessica dai @jessicadai_]

im no expert but seems like a phd is all about having a bad time and reading a lot of brandon taylor and then later figuring out how to psy op urself into having a good time!

[Gothamist @Gothamist]

Mayor Eric Adams told at least one member of the City Council he would restore cuts to their favorite programs if they agreed to vote against a police transparency bill the mayor and NYPD officials vigorously opposed. https://t.co/wQX2n6y11c

[Gabriel Agostini @gsagostini]

We apply the method to infer flooding probabilities following the passage of Hurricane Ida in NYC, one of the deadliest and most expensive storms hitting the Northeast this past decade.

And we make cool maps!🗺️6/8

[Sidhika Balachandar @sidhikab1]

Check out our new paper, which we’ll present as a spotlight talk at NeurIPS DistShift. This is joint work with @NikhGarg and @2plus2make5. 1/N

[Emma Pierson @2plus2make5]

Our paper measuring socioeconomic segregation from cell phone mobility data is out in @Nature today!

[Emma Pierson @2plus2make5]

Excited to announce the our new Coursera course, "Practical Steps for Building Fair Algorithms", taught with @__kkado and developed with @jtleek and @fredhutch (feel free to retweet)

This class is meant for everyone! It is free to audit and doesn't require CS background.

[Leah Pierson @leah_pierson]

Some research projects are vastly and predictably more socially valuable than others, and researchers should choose the more socially valuable ones.