ListDB RepThree: Stationary camera videos and trajectories at a signalized 3-way intersection taken on different Wednesdays over a 3-year period (2019, 2020 with Covid-19 lockdown, 2022)

Abstract

The dataset can be used to study road traffic before, during, and after Covid-19. Moreover, the dataset can help to derive test scenarios for scenario-based testing, for example.

The dataset includes seven days of drone videos and their corresponding trajectories recorded in Dresden, Germany, at a signalized 3-way intersection. Data collection in 2020 also took place on days with Covid-19 conditional lockdown, including curfew. The data collection followed a fixed sampling plan covering four different time slots daily (weekdays: 07.30-09.00 / 10.00-11.30 / 13.00-14.30 / 15.30-17.00) within 2019-2022. Moreover, extensive metadata is provided following the ListDB specification (w3id.org/listdb). Source trajectory data was extracted from video using DataFromSky TrafficSurvey (https://ai.datafromsky.com/aerial) - an AI video analytics-based service for gathering advanced traffic data for researchers and traffic engineers. Main facts: 1056 minutes of video (2019: 160 minutes / 2020: 560 minutes / 2022: 336 minutes). The analysed trajectories are available for each video. We sincerely thank DataFromSky for supporting the analysis. The video-based traffic observation was conducted within the MOTUS project (19FS2015A; motus-project.com) and supported by the mFUND grant by Germany’s Federal Ministry for Digital and Transport.