Vision-Based Extraction of Vehicle and Pedestrian Trajectories for Behavioral Analysis of Street Interventions

Institute

Lehrstuhl für Fahrzeugtechnik (TUM-ED)

Type

Semester Thesis / Master's Thesis /

Content

experimental / theoretical /  

Description

As part of MCube MOSAIQ (Mobility and Urban Climate in Future Neighborhoods), TUM together with the City of Munich and partners is temporarily redesigning street sections and public spaces in Schwabing-West and Moosach, adding gathering spaces, greenery, safer crossings, and cyclist amenities, and scientifically evaluating how these interventions change everyday mobility behavior.

A central question is whether people actually change their behavior after an intervention, in particular where they choose to stop, linger, and spend time in the redesigned space. Cameras have been installed at multiple intervention locations to record street-level footage.

This thesis focuses on building a machine-vision pipeline that detects and tracks vehicles and pedestrians across the recorded footage and turns it into a clean, structured trajectory database, which will form the basis for downstream behavioral analyses.

Requirements

Programming skills in Python
Interest in computer vision / deep learning (e.g. object detection, tracking)
Independent and structured working style

Tags

FTM Studienarbeit, FTM SM, FTM Zacher, FTM Informatik

Possible start

sofort

Contact

Till Zacher, M.Sc.
Phone: +49 89 289 15351
till.zachertum.de

Announcement