Computer Vision Traffic Sensing UROP

Stuck in traffic with computer vision

Interested in a cutting-edge application of computer vision algorithms with social impact? Do you want to hone your ML skills while studying an important urban problem?

Background: Jakarta, Indonesia, is one of the cities with the worst traffic congestion in the world. On some routes, peak-time speeds crawl below 6 miles/hour Every. Single. Day.

Project. This project will collect and process real-time video feeds from thousands of cameras across Jakarta. The goal is to construct a detailed dataset of vehicle counts and types, by road segment.

This project will feed into a study of the public transport network and its impact on traffic congestion.

Challenge. The computer vision challenge is to train accurate, scalable deep neural network classifiers that can be applied in near-real time on hundreds of cameras.

You. We seek an exceptional summer full-time UROP to make an important contribution to this project. You will work closely with researchers and faculty at U Chicago, MIT and Harvard, to develop and launch this data collection effort.

The ideal candidate:

• Computer science major or extensive related course and project experience
• Familiarity with computer vision algorithms (e.g. YOLO, Mask-RCNN)
• Proficiency with python, command line
• Keen attention to detail and perseverance

Position details:

• Paid position.
• Full-time during summer.

To apply, send an email with subject line “[RA computer vision]” with a CV and a short paragraph with your experience with ML (courses and projects), and an unofficial transcript.

Contact Name: Gabriel Kreindler, Research Fellow, Becker-Friedman Institute

Contact Email: gekr@uchicago.edu