This dataset is used for the classification of traffic intersection regulations using (car) GPS trajectory data. The coverage is mostly a large part of the city of Hannover, Germany.

The overview of the GPS trajectory dataset is given in the below figure: https://data.uni-hannover.de/dataset/379ed322-a9ea-48f3-bc13-2f5ea3174470/resource/ecfbc059-9ca3-4cbc-8774-de091f0fbcd6/download/hannover_traj.png" alt="Overview">

When the trajectory dataset is combined with the related intersection ground-truth information (available at: https://doi.org/10.25835/cqg0x1el): https://data.uni-hannover.de/dataset/379ed322-a9ea-48f3-bc13-2f5ea3174470/resource/bc832922-aedd-4bac-86b3-5d4a753bfecc/download/hannover_rules_traj.png" alt="Combined">

Data Acquisition

The trajectory samples were recorded using an android smartphone while driving a car in and around the city of Hannover, Germany. The acquisition period was from December 2017 to March 2019 by only a single person. The recording of the trajectories has taken place without restrictions in order to reflect a normal behavior of everyday car journeys. The sampling rate is approximately 1 sample per second.

Related Publications:

• Zourlidou, S., Sester, M. and Hu, S. (2022): Recognition of Intersection Traffic Regulations From Crowdsourced Data. Preprints 2022, 2022070012. DOI: https://doi.org/10.20944/preprints202207.0012.v1

• Zourlidou, S., Golze, J. and Sester, M. (2022): Traffic Regulation Recognition using Crowd-Sensed GPS and Map Data: a Hybrid Approach, AGILE GIScience Ser., 3, 22, 2022. https://doi.org/10.5194/agile-giss-3-22-2022

• Cheng, H., Lei, H., Zourlidou, S., Sester, M. (2022): Traffic Control Recognition with an Attention Mechanism Using Speed-Profile and Satellite Imagery data. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences XLIII-B4-2022, S. 287–29. https://doi.org/10.5194/isprs-archives-XLIII-B4-2022-287-2022

• Wang, C., Zourlidou, S., Golze, J. and Sester, M. (2020): Trajectory analysis at intersections for traffic rule identification. Geo-spatial Information Science, 11(4):1-10. https://doi.org/10.1080/10095020.2020.1843374

• Cheng, H., Zourlidou, S. and Sester, M. (2020): Traffic Control Recognition with Speed-Profiles: A Deep Learning Approach. ISPRS Int. J. Geo-Inf. 2020, 9, 652. https://doi.org/10.3390/ijgi9110652

• Golze, J., Zourlidou, S. and Sester, M. (2020): Traffic Regulator Detection Using GPS Trajectories. KN J. Cartogr. Geogr. Inf. https://doi.org/10.1007/s42489-020-00048-x

• Zourlidou, S., Fischer, C. and Sester, M. (2019): Classification of street junctions according to traffic regulators. In: Kyriakidis, P., Hadjimitsis, D., Skarlatos, D. and Mansourian, A., (eds) 2019. Accepted short papers and posters from the 22nd AGILE conference on geo-information science. Cyprus University of Technology 17–20 June 2019, Limassol, Cyprus.

Related Datasets:

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: Traffic Regulator Ground-truth Information for the Chicago Trajectory Dataset. https://doi.org/10.25835/0vifyzqi

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: GPS Trajectory Dataset and Traffic Regulation Information of the Region of Edessa, Greece. https://doi.org/10.25835/v0mzwob3

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: Traffic Regulator Ground-truth Information of the City of Hannover, Germany. https://doi.org/10.25835/cqg0x1el

• Zourlidou, S., Golze, J. and Sester, M. (2020). Dataset: Speed profiles and GPS Trajectories for Traffic Rule Recognition (6 Junctions, Hannover, Germany). https://doi.org/10.25835/0043786

The Microsoft Trajectory Dataset is a collection of data that includes information about the movement trajectories of various objects in the physical world. It is designed for use in research and development of applications related to tracking, motion analysis, and prediction. The dataset contains time-stamped trajectory data captured from different sources, such as GPS devices, mobile phones, and sensors. The objects whose trajectories are included in the dataset can range from vehicles, pedestrians, and bicycles to animals and other moving objects. The trajectory data typically includes coordinates (e.g., latitude and longitude), velocity, and direction of movement for each object at different points in time. The dataset may also include additional information such as object type, timestamp accuracy, and contextual information such as weather conditions, road network, or other environmental factors.The Microsoft Trajectory Dataset is large and diverse, covering a wide range of scenarios, such as urban environments, rural areas, and indoor spaces. It can be used for various applications, such as transportation planning, traffic management, urban mobility analysis, pedestrian behavior modeling, and prediction of object movement patterns. Researchers and developers can leverage the Microsoft Trajectory Dataset to develop and evaluate algorithms, models, and systems related to trajectory prediction, object tracking, and motion analysis. The dataset provides a valuable resource for studying and understanding the dynamics of object movement in the real world, and it can be used to advance the field of trajectory analysis and prediction.

OSM point features in Europe with tags categorized as POI in the Geofabrik taxonomy. Extracted from original OpenStreetMap database (osm.pbf) in 09.2019 (https://www.openstreetmap.org/#map=7/46.825/8.224)

This dataset is a subset of a much larger data collection and used for the analysis of speed- and time-profiles of trajectories crossing different selected intersections. Resulting findings can be used for the intersection categorization according to traffic regulation types.

The six selected intersections (A - F) and the crossing trajectory samples (green) can be seen in the subsequent figure: https://data.uni-hannover.de/dataset/1a97c0df-4659-43bd-99e9-d2573d31f5cd/resource/0571e0e3-19d6-4e45-9da5-6cb932a3d7d7/download/junctionall.png" alt="Intersections">

Data Acquisition

The trajectory samples were recorded using an android smartphone while driving a car in and around the city of Hannover, Germany. The acquisition period was from December 2017 to March 2019 by only a single person. The recording of the trajectories has taken place without restrictions in order to reflect a normal behavior of everyday car journeys. The sampling rate is approximately 1 sample per second.

Related Publications:

• Zourlidou, S., Sester, M. and Hu, S. (2022): Recognition of Intersection Traffic Regulations From Crowdsourced Data. Preprints 2022, 2022070012. DOI: https://doi.org/10.20944/preprints202207.0012.v1

• Zourlidou, S., Golze, J. and Sester, M. (2022): Traffic Regulation Recognition using Crowd-Sensed GPS and Map Data: a Hybrid Approach, AGILE GIScience Ser., 3, 22, 2022. https://doi.org/10.5194/agile-giss-3-22-2022

• Cheng, H., Lei, H., Zourlidou, S., Sester, M. (2022): Traffic Control Recognition with an Attention Mechanism Using Speed-Profile and Satellite Imagery data. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences XLIII-B4-2022, S. 287–29. https://doi.org/10.5194/isprs-archives-XLIII-B4-2022-287-2022

• Wang, C., Zourlidou, S., Golze, J. and Sester, M. (2020): Trajectory analysis at intersections for traffic rule identification. Geo-spatial Information Science, 11(4):1-10. https://doi.org/10.1080/10095020.2020.1843374

• Cheng, H., Zourlidou, S. and Sester, M. (2020): Traffic Control Recognition with Speed-Profiles: A Deep Learning Approach. ISPRS Int. J. Geo-Inf. 2020, 9, 652. https://doi.org/10.3390/ijgi9110652

• Golze, J., Zourlidou, S. and Sester, M. (2020): Traffic Regulator Detection Using GPS Trajectories. KN J. Cartogr. Geogr. Inf. https://doi.org/10.1007/s42489-020-00048-x

• Zourlidou, S., Fischer, C. and Sester, M. (2019): Classification of street junctions according to traffic regulators. In: Kyriakidis, P., Hadjimitsis, D., Skarlatos, D. and Mansourian, A., (eds) 2019. Accepted short papers and posters from the 22nd AGILE conference on geo-information science. Cyprus University of Technology 17–20 June 2019, Limassol, Cyprus.

Related Datasets:

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: GPS Trajectory Dataset of the Region of Hannover, Germany. https://doi.org/10.25835/9bidqxvl

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: Traffic Regulator Ground-truth Information for the Chicago Trajectory Dataset. https://doi.org/10.25835/0vifyzqi

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: GPS Trajectory Dataset and Traffic Regulation Information of the Region of Edessa, Greece. https://doi.org/10.25835/v0mzwob3

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: Traffic Regulator Ground-truth Information of the City of Hannover, Germany. https://doi.org/10.25835/cqg0x1el

This dataset is used for the classification of traffic intersections regulation rules using car GPS trajectories. In order to apply supervised classification methods, the ground-truth information is also contained in this dataset. The regulators are annotated based on the intersection arms at each intersection.

An overview of the dataset's trajectories can be seen in the subsequent figure: https://data.uni-hannover.de/dataset/57869009-330e-49be-bc63-56fb56952bdb/resource/c5fddb86-82f5-491a-9099-51e5a9109cda/download/edessa_traj.png" alt="Overview of the dataset containing the trajectories collected in Edessa.">

An overview of the related intersection regulator ground-truth rules can be seen in the following figure: https://data.uni-hannover.de/dataset/57869009-330e-49be-bc63-56fb56952bdb/resource/c8d75ea3-83ed-445a-a1db-38ca092f94a6/download/edessa_traj_junc_rules.png" alt="">

Data Acquisition

The trajectory samples were recorded using an android smartphone while driving a car in and around the city of Edessa, Greece. The acquisition period was from March 2018 to September 2018 (6 month) by only a single person. The recording of the trajectories has taken place without restrictions in order to reflect a normal behavior of everyday car journeys. The sampling rate is approximately 1 sample per second. Additionally to the GPS trajectories, the ground-truth regulator rules of the traffic intersections were annotated via street view images.

Related Publications:

• Zourlidou, S., Sester, M. and Hu, S. (2022): Recognition of Intersection Traffic Regulations From Crowdsourced Data. Preprints 2022, 2022070012. DOI: https://doi.org/10.20944/preprints202207.0012.v1

• Zourlidou, S., Golze, J. and Sester, M. (2022): Traffic Regulation Recognition using Crowd-Sensed GPS and Map Data: a Hybrid Approach, AGILE GIScience Ser., 3, 22, 2022. https://doi.org/10.5194/agile-giss-3-22-2022

• Cheng, H., Lei, H., Zourlidou, S., Sester, M. (2022): Traffic Control Recognition with an Attention Mechanism Using Speed-Profile and Satellite Imagery data. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences XLIII-B4-2022, S. 287–29. https://doi.org/10.5194/isprs-archives-XLIII-B4-2022-287-2022

• Wang, C., Zourlidou, S., Golze, J. and Sester, M. (2020): Trajectory analysis at intersections for traffic rule identification. Geo-spatial Information Science, 11(4):1-10. https://doi.org/10.1080/10095020.2020.1843374

• Cheng, H., Zourlidou, S. and Sester, M. (2020): Traffic Control Recognition with Speed-Profiles: A Deep Learning Approach. ISPRS Int. J. Geo-Inf. 2020, 9, 652. https://doi.org/10.3390/ijgi9110652

• Golze, J., Zourlidou, S. and Sester, M. (2020): Traffic Regulator Detection Using GPS Trajectories. KN J. Cartogr. Geogr. Inf. https://doi.org/10.1007/s42489-020-00048-x

• Zourlidou, S., Fischer, C. and Sester, M. (2019): Classification of street junctions according to traffic regulators. In: Kyriakidis, P., Hadjimitsis, D., Skarlatos, D. and Mansourian, A., (eds) 2019. Accepted short papers and posters from the 22nd AGILE conference on geo-information science. Cyprus University of Technology 17–20 June 2019, Limassol, Cyprus.

Related Datasets:

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: GPS Trajectory Dataset of the Region of Hannover, Germany. https://doi.org/10.25835/9bidqxvl

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: Traffic Regulator Ground-truth Information for the Chicago Trajectory Dataset. https://doi.org/10.25835/0vifyzqi

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: Traffic Regulator Ground-truth Information of the City of Hannover, Germany. https://doi.org/10.25835/cqg0x1el

• Zourlidou, S., Golze, J. and Sester, M. (2020). Dataset: Speed profiles and GPS Trajectories for Traffic Rule Recognition (6 Junctions, Hannover, Germany). https://doi.org/10.25835/0043786

CATS: Conditional Adversarial Trajectory Synthesis for Privacy-Preserving Trajectory Data Publication Using Deep Learning ApproachesJinmeng Rao, Song Gao , Sijia ZhuGeoDS Lab, Department of Geography, University of Wisconsin-Madison, WIInternational Journal of Geographical Information ScienceAbstract: The prevalence of ubiquitous location-aware devices and mobile Internet enables us to collect massive individual-level trajectory dataset from users. Such trajectory big data bring new opportunities to human mobility and GeoAI research but also raise public concerns with regard to location privacy. In this work, we present the Conditional Adversarial Trajectory Synthesis (CATS), a deep-learning-based methodological framework for privacy-preserving trajectory data publication. CATS applies K-anonymity to the underlying spatiotemporal distributions of aggregated human mobility, which provides a distributional-level strong privacy guarantee. By leveraging conditional adversarial training on K-anonymized human mobility matrices, trajectory global context learning using the attention-based mechanism, and recurrent bipartite graph matching of adjacent trajectory points, CATS is able to reconstruct trajectory topology from conditionally sampled locations and generate high-quality synthetic trajectory data, which can serve as supplements or alternatives to raw data for privacy-preserving trajectory data publication. The experiment results on over 90k trajectories show that our method has a better performance in privacy preservation, characteristic preservation, and downstream utility compared with baseline methods, which brings new insights into privacy-preserving human mobility research and explores data ethics issues in GIScience.Source code for the work entitled "CATS: Conditional Adversarial Trajectory Synthesis for Privacy-Preserving Trajectory Data Publication Using Deep Learning Approaches".Due to the non-disclosure agreement with the data provider, we are not releasing the original individual-level GPS trajectory data but sharing the mocked individual GPS trajectory data with the same data structure and also the k-anonymized aggregated human mobility data used in the experiments.Content Introduction- aggregate_mobility_matrix.py: generate the aggregated mobility distribution file from individual GPS trajectories- cats.py: provides a PyTorch implementation of CATS, including a TrajGenerator (CatGen) class and a TrajDiscriminator (CatCrt) class.- train_cats.py: provides a training example code for CATS.- run_cats.py: provides a inference example code for CATS.- log_util.py: logging class.- stmm_dataset.py: provides a Dataset class used for training the CATS.- stmm_data/: k-anonymized aggregated mobility matrix data.- mocked_individual_gps_data: the mocked individual GPS trajectory data samples with the same data structure with raw data

Overview

This dataset provides high-resolution, georeferenced vehicle trajectories collected via drone footage at three roundabouts located in the municipalities of Frick and Laufenburg, Canton of Aargau, Switzerland. The data were collected as part of a collaborative drone campaign organized by the Urban Transport Systems Laboratory (LUTS), EPFL, within the framework of NCCR Automation, in cooperation with the cantonal traffic planning department of Aargau. The collection took place on Monday, 23rd October 2023, during peak morning and afternoon hours, resulting in nearly 11 hours of 4K video data.

Dataset Composition

This dataset contains CSV files structured with consistent data fields representing georeferenced trajectories, vehicle types (car, bus, truck), and timestamps, capturing detailed vehicle movements within roundabout environments.

File Organization

File names follow the convention:

D{X}_{TP}{N}_{S}.csv

D{X} — the drone identifier, where {X} is a number (e.g., 1, 2) indicating which drone captured the data.→ Example: D1 = data collected by Drone 1.

{TP}{N} — the time period and session number, where {TP} is either AM (morning) or PM (afternoon), and {N} is an integer indicating the session number.→ Example: AM2 = second morning session.

{S} — the site identifier, corresponding to one of the monitored sites:→ F1 = Roundabout F1 (Frick)→ F2 = Roundabout F2 (Frick)→ L1 = Roundabout L1 (Laufenburg)

CSV File Structure

Each CSV file includes:

Column Name Description Format / Units

track_id Unique vehicle identifier (per file) Integer

type Vehicle type (Car, Bus, Truck) Categorical

lon WGS84 geographic longitude Decimal degrees (15 d.p.)

lat WGS84 geographic latitude Decimal degrees (15 d.p.)

time Local timestamp in ISO 8601 format String (hh:mm:ss.ss)

Data Collection and Processing

Collection Method: Two drones flying at an altitude of 120 meters above ground level, capturing videos at 4K resolution (3840×2160 pixels) at 29.97 FPS.

Locations:

Roundabout F1 (Frick): Intersection of Bahnhofstrasse and Hauptstrasse 3 (Urban)

Roundabout F2 (Frick): Intersection of Hauptstrasse 3 with Gänsacker and Stöcklimattstrasse (Urban)

Roundabout L1 (Laufenburg): Intersection at Hauptstrasse 7 near the German border (Rural)

Data Processing: The detection, tracking, and trajectory stabilization were performed using the early version of the Geo-trax framework (v0.1.0), an advanced computer vision pipeline tailored for drone-captured traffic footage. The resulting trajectories are precisely represented in stabilized pixel coordinates, which are subsequently transformed into geographic coordinates (WGS84). This georeferencing process follows a procedure similar to that described in Espadaler-Clapés et al., 2023, and includes:

Identification and extraction of Ground Control Points (GCPs) in the first stabilized video frame using QGIS Georeferencer, linking pixel coordinates to UTM coordinates.

Linear regression modeling between stabilized pixel coordinates and corresponding UTM coordinates derived from GCPs to estimate transformation parameters.

Projection to WGS84, converting UTM coordinates into global geographic coordinates using a standard GIS transformation (EPSG:4326).

Dataset Statistics

Roundabout Videos Avg. Duration (min) Total Duration (min) Vehicles (total) Cars Buses Trucks

F1 8 18.63 149.04 4,283 3,967 72 244

F2 6 19.24 115.44 2,528 2,205 26 297

L1 4 20.39 81.56 2,130 1,980 24 126

Potential Applications

This dataset is well-suited for:

Gap acceptance behavior studies at roundabouts (e.g., Pascual Anglès et al., 2025)

Traffic flow analysis and modeling

Safety assessments using surrogate safety measures (SSMs)

Validation of traffic simulation models

A trajectory restoration algorithm for low-sampling-rate floating car data and complex urban road networks

This dataset contains the ground-truth intersection regulators for all intersections crossed by the open GPS trajectory data of Chicago, published on Mapconstruction.org.

Data Acquisition

The ground-truth data is generated by manually inspecting images at the intersections crossed by the published trajectory dataset on Mapconstruction.org. These images are derived from mapillary.com at the respective intersections. Satellite images are investigated as an additional data source, if needed. This procedure result in labels for each intersection arm.

Related Publications:

• Zourlidou, S., Sester, M. and Hu, S. (2022): Recognition of Intersection Traffic Regulations From Crowdsourced Data. Preprints 2022, 2022070012. DOI: https://doi.org/10.20944/preprints202207.0012.v1

• Zourlidou, S., Golze, J. and Sester, M. (2022): Traffic Regulation Recognition using Crowd-Sensed GPS and Map Data: a Hybrid Approach, AGILE GIScience Ser., 3, 22, 2022. https://doi.org/10.5194/agile-giss-3-22-2022

• Cheng, H., Lei, H., Zourlidou, S., Sester, M. (2022): Traffic Control Recognition with an Attention Mechanism Using Speed-Profile and Satellite Imagery data. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences XLIII-B4-2022, S. 287–29. https://doi.org/10.5194/isprs-archives-XLIII-B4-2022-287-2022

• Wang, C., Zourlidou, S., Golze, J. and Sester, M. (2020): Trajectory analysis at intersections for traffic rule identification. Geo-spatial Information Science, 11(4):1-10. https://doi.org/10.1080/10095020.2020.1843374

• Cheng, H., Zourlidou, S. and Sester, M. (2020): Traffic Control Recognition with Speed-Profiles: A Deep Learning Approach. ISPRS Int. J. Geo-Inf. 2020, 9, 652. https://doi.org/10.3390/ijgi9110652

• Golze, J., Zourlidou, S. and Sester, M. (2020): Traffic Regulator Detection Using GPS Trajectories. KN J. Cartogr. Geogr. Inf. https://doi.org/10.1007/s42489-020-00048-x

• Zourlidou, S., Fischer, C. and Sester, M. (2019): Classification of street junctions according to traffic regulators. In: Kyriakidis, P., Hadjimitsis, D., Skarlatos, D. and Mansourian, A., (eds) 2019. Accepted short papers and posters from the 22nd AGILE conference on geo-information science. Cyprus University of Technology 17–20 June 2019, Limassol, Cyprus.

Related Datasets:

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: GPS Trajectory Dataset of the Region of Hannover, Germany. https://doi.org/10.25835/9bidqxvl

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: GPS Trajectory Dataset and Traffic Regulation Information of the Region of Edessa, Greece. https://doi.org/10.25835/v0mzwob3

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: Traffic Regulator Ground-truth Information of the City of Hannover, Germany. https://doi.org/10.25835/cqg0x1el

• Zourlidou, S., Golze, J. and Sester, M. (2020). Dataset: Speed profiles and GPS Trajectories for Traffic Rule Recognition (6 Junctions, Hannover, Germany). https://doi.org/10.25835/0043786

Overview

The Songdo Traffic dataset delivers precisely georeferenced vehicle trajectories captured through high-altitude bird's-eye view (BeV) drone footage over Songdo International Business District, South Korea. Comprising approximately 700,000 unique trajectories, this resource represents one of the most extensive aerial traffic datasets publicly available, distinguishing itself through exceptional temporal resolution that captures vehicle movements at 29.97 points per second, enabling unprecedented granularity for advanced urban mobility analysis.

⚠️ Important: If you use this dataset in your work, please cite the following reference [1]:

Robert Fonod, Haechan Cho, Hwasoo Yeo, Nikolas Geroliminis (2025). Advanced computer vision for extracting georeferenced vehicle trajectories from drone imagery, arXiv preprint arXiv:2411.02136.

(Note: This manuscript shall be replaced by the published version once available.)

Dataset Composition

The dataset consists of four primary components:

Trajectory Data: 80 ZIP archives containing high-resolution vehicle trajectories with georeferenced positions, speeds and acceleration profiles, and other metadata.

Orthophoto Cut-Outs: High-resolution (8000×8000 pixel) orthophoto images for each monitored intersection, used for georeferencing and visualization.

Road and Lane Segmentations: CSV files defining lane polygons within road sections, facilitating mapping of vehicle positions to road segments and lanes.

Sample Videos: A selection of 4K UHD drone video samples capturing intersection footage during the experiment.

Data Collection

The dataset was collected as part of a collaborative multi-drone experiment conducted by KAIST and EPFL in Songdo, South Korea, from October 4–7, 2022.

A fleet of 10 drones monitored 20 busy intersections, executing advanced flight plans to optimize coverage.

4K (3840×2160) RGB video footage was recorded at 29.97 FPS from altitudes of 140–150 meters.

Each drone flew 10 sessions per day, covering peak morning and afternoon periods.

The experiment resulted in 12TB of 4K raw video data.

More details on the experimental setup and data processing pipeline are available in [1].

Data Processing

The trajectories were extracted using geo-trax, an advanced deep learning framework designed for high-altitude UAV-based traffic monitoring. This state-of-the-art pipeline integrates vehicle detection, tracking, trajectory stabilization, and georeferencing to extract high-accuracy traffic data from drone footage.

Key Processing Steps:

Vehicle Detection & Tracking: Vehicles were detected and tracked across frames using a deep learning-based detector and motion-model-based tracking algorithm.

Trajectory Stabilization: A novel track stabilization method was applied using detected vehicle bounding boxes as exclusion masks in image registration.

Georeferencing & Coordinate Transformation: Each trajectory was transformed into global (WGS84), local Cartesian, and orthophoto coordinate systems.

Vehicle Metadata Estimation: In addition to time-stamped vehicle trajectories, various metadata attributes were also extracted, including vehicle dimensions and type, speed, acceleration, class, lane number, road section, and visibility status.

More details on the extraction methodology are available in [1].

File Structure & Formats

1. Trajectory Data (Daily Intersection ZIPs, 16.2 MB ~ 360.2 MB)

The trajectory data is organized into 80 ZIP files, each containing traffic data for a specific intersection and day of the experiment.

File Naming Convention:

YYYY-MM-DD_intersectionID.zip

YYYY-MM-DD represents the date of data collection (2022-10-04 to 2022-10-07).

intersectionID is a unique identifier for one of the 20 intersections where data was collected (A, B, C, E, …, U). The letter D is reserved to denote "Drone".

Each ZIP file contains 10 CSV files, each corresponding to an individual flight session:

YYYY-MM-DD_intersectionID.zip │── YYYY-MM-DD_intersectionID_AM1.csv ├── … │── YYYY-MM-DD_intersectionID_AM5.csv │── YYYY-MM-DD_intersectionID_PM1.csv ├── … └── YYYY-MM-DD_intersectionID_PM5.csv

Here, AM1-AM5 and PM1-PM5 denote morning and afternoon flight sessions, respectively. For example, 2022-10-04_S_AM1.csv contains all extracted trajectories from the first morning session of the first day at the intersection 'S'.

CSV File Example Structure:

Each CSV file contains high-frequency trajectory data, formatted as follows (d.p. = decimal place):

Dataset Column Name Format / Units Data Type Explanation

Vehicle_ID 1, 2, … Integer Unique vehicle identifier within each CSV file

Local_Time hh:mm:ss.sss String Local Korean time (GMT+9) in ISO 8601 format

Drone_ID 1, 2, …, 10 Integer Unique identifier for the drone capturing the data

Ortho_X, Ortho_Y px (1 d.p.) Float Vehicle center coordinates in the orthophoto cut-out image

Local_X, Local_Y m (2 d.p.) Float KGD2002 / Central Belt 2010 planar coordinates (EPSG:5186)

Latitude, Longitude ° DD (7 d.p.) Float WGS84 geographic coordinates in decimal degrees (EPSG:4326)

Vehicle_Length*, Vehicle_Width* m (2 d.p.) Float Estimated physical dimensions of the vehicle

Vehicle_Class Categorical (0–3) Integer Vehicle type: 0 (car/van), 1 (bus), 2 (truck), 3 (motorcycle)

Vehicle_Speed* km/h (1 d.p.) Float Estimated speed computed from trajectory data using Gaussian smoothing

Vehicle_Acceleration* m/s² (2 d.p.) Float Estimated acceleration derived from smoothed speed values

Road_Section* N_G String Road section identifier (N = node, G = lane group)

Lane_Number* 1, 2, … Integer Lane position (1 = leftmost lane in the direction of travel)

Visibility 0/1 Boolean 1 = fully visible, 0 = partially visible in the camera frame

• These columns may be empty under certain conditions, see [1] for more details.

1. Orthophoto Cut-Outs (orthophotos.zip, 1.8 GB)

For each intersection, we provide the high-resolution orthophoto cut-outs that were used for georeferencing. These 8000×8000 pixel PNG images cover specific areas, allowing users to overlay orthophoto trajectories within the road network.

orthophotos/ │── A.png │── B.png │── … └── U.png

For more details on the orthophoto generation process, refer to [1].

1. Orthophoto Segmentations (segmentations.zip, 24.9 KB)

We provide the road and lane segmentations for each orthophoto cut-out, stored as CSV files where each row defines a lane polygon within a road section.

Each section (N_G) groups lanes moving in the same direction, with lanes numbered sequentially from the innermost outward. The CSV files are structured as follows:

segmentations/ │── A.csv │── B.csv │── … └── U.csv

Each file contains the following columns:

Section: Road section ID (N_G format).

Lane: Lane number within the section.

tlx, tly, blx, bly, brx, bry, trx, try: Polygon corner coordinates.

These segmentations enabled trajectory points to be mapped to specific lanes and sections in our trajectory dataset. Vehicles outside segmented areas (e.g., intersection centers) remain unlabeled. Perspective distortions may also cause misalignments for taller vehicles.

1. Sample Videos (sample_videos.zip, 26.8 GB)

The dataset includes 29 video samples, each capturing the first 60 seconds of drone hovering over its designated intersection during the final session (PM5) on October 7, 2022. These high-resolution 4K videos provide additional context for trajectory analysis and visualization, complementing the orthophoto cut-outs and segmentations.

sample_videos/ │── A_D1_2022-10-07_PM5_60s.mp4 │── A_D2_2022-10-07_PM5_60s.mp4 │── B_D1_2022-10-07_PM5_60s.mp4 │── … └── U_D10_2022-10-07_PM5_60s.mp4

Additional Files

README.md – Dataset documentation (this file)

LICENSE.txt – Creative Commons Attribution 4.0 License

Known Dataset Artifacts and Limitations

While this dataset is designed for high accuracy, users should be aware of the following known artifacts and limitations:

Trajectory Fragmentation: Trajectories may be fragmented for motorcycles in complex road infrastructure scenarios (pedestrian crossings, bicycle lanes, traffic signals) and for certain underrepresented truck variants. Additional fragmentations occurred when drones experienced technical issues during hovering, necessitating mid-recording splits that naturally resulted in divided trajectories.

Vehicle ID Ambiguities: The largest Vehicle_ID in a CSV file does not necessarily indicate the total number of unique vehicles.

Kinematic Estimation Limitations: Speed and acceleration values are derived from raw tracking data and may be affected by minor errors due to detection inaccuracies, stabilization artifacts, and applied interpolation and smoothing techniques.

Vehicle Dimension Estimation: Estimates may be unreliable for stationary or non-axially moving vehicles and can be affected by bounding box overestimations capturing protruding vehicle parts or shadows.

Lane and Section Assignment Inaccuracies: Perspective effects may cause vehicles with significant heights, such as trucks or buses, to be misassigned to incorrect lanes or sections in the orthophoto.

Occasional pedestrian pair misclassifications: Rarely, two pedestrians walking side by side may be briefly mistaken for a motorcycle, but such instances are short-lived and typically removed by the short trajectory filter.

For a comprehensive discussion of dataset limitations and validation procedures, refer to [1].

Citation & Attribution

Preferred Citation:

If you use Songdo Traffic for any purpose, whether in academic research, commercial applications, open-source projects, or benchmarking efforts, please cite our accompanying manuscript [1]:

Robert Fonod, Haechan Cho, Hwasoo Yeo, Nikolas Geroliminis (2025). Advanced computer vision for extracting georeferenced vehicle

Interesting sequential patterns in human movement trajectories can provide valuable knowledge for urban management, planning, and location-based business. Existing methods for mining such patterns, however, tend not to consider the reduced likeliness of trips with increasing travel cost. Consequently, it is difficult to differentiate the patterns emerging from people’s specific travel interests from those simply due to travel convenience. To solve this problem, this article presents Geo-SigSPM for mining geographically interesting and statistically significant sequential patterns from trajectories. Here, ‘geographically interesting’ patterns are those more frequent than their expected frequencies which consider both the travel cost and non-redundancy of any place in the patterns. To achieve this, Geo-SigSPM formulates the expected frequencies of the patterns based on doubly-constrained human mobility models and the frequencies of their subsequences. A set of statistical tests is also developed to evaluate the identified interesting patterns. Experiments with synthetic and Foursquare check-in datasets demonstrate the efficacy of Geo-SigSPM in discovering geographically interesting patterns, controlling the spurious pattern rate, and discovering patterns that better reflect people’s specific travel interests than the conventional frequency-based pattern mining approach. Geo-SigSPM is a promising solution to improving relevant decision-making when people’s travel preference beyond travel cost is concerned.

Context

Technology has many effects on the transportation industry.

Content

We have provided an accurate dataset describing a complete year (from 01/07/2013 to 30/06/2014) of the trajectories for all the 442 taxis running in the city of Porto, in Portugal (i.e. one CSV file named "train.csv"). These taxis operate through a taxi dispatch central, using mobile data terminals installed in the vehicles. We categorize each ride into three categories: A) taxi central based, B) stand-based or C) non-taxi central based. For the first, we provide an anonymized id, when such information is available from the telephone call. The last two categories refer to services that were demanded directly to the taxi drivers on a B) taxi stand or on a C) random street.

Each data sample corresponds to one completed trip. It contains a total of 9 (nine) features, described as follows:

• TRIP_ID: (String) It contains an unique identifier for each trip;

• CALL_TYPE: (char) It identifies the way used to demand this service. It may contain one of three possible values: ‘A’ if this trip was dispatched from the central; ‘B’ if this trip was demanded directly to a taxi driver on a specific stand; ‘C’ otherwise (i.e. a trip demanded on a random street).

• ORIGIN_CALL: (integer) It contains an unique identifier for each phone number which was used to demand, at least, one service. It identifies the trip’s customer if CALL_TYPE=’A’. Otherwise, it assumes a NULL value;

• ORIGIN_STAND: (integer): It contains an unique identifier for the taxi stand. It identifies the starting point of the trip if CALL_TYPE=’B’. Otherwise, it assumes a NULL value;

• TAXI_ID: (integer): It contains an unique identifier for the taxi driver that performed each trip;

• TIMESTAMP: (integer) Unix Timestamp (in seconds). It identifies the trip’s start;

• DAYTYPE: (char) It identifies the daytype of the trip’s start. It assumes one of three possible values: ‘B’ if this trip started on a holiday or any other special day (i.e. extending holidays, floating holidays, etc.); ‘C’ if the trip started on a day before a type-B day; ‘A’ otherwise (i.e. a normal day, workday or weekend).

• MISSING_DATA: (Boolean) It is FALSE when the GPS data stream is complete and TRUE whenever one (or more) locations are missing

• POLYLINE: (String): It contains a list of GPS coordinates (i.e. WGS84 format) mapped as a string. The beginning and the end of the string are identified with brackets (i.e. [ and ], respectively). Each pair of coordinates is also identified by the same brackets as [LONGITUDE, LATITUDE]. This list contains one pair of coordinates for each 15 seconds of trip. The last list item corresponds to the trip’s destination while the first one represents its start;

The total travel time of the trip (the prediction target of this competition) is defined as the (number of points-1) x 15 seconds. For example, a trip with 101 data points in POLYLINE has a length of (101-1) * 15 = 1500 seconds. Some trips have missing data points in POLYLINE, indicated by MISSING_DATA column, and it is part of the challenge how you utilize this knowledge.

Acknowledgements

Data from ECML/PKDD 15: Taxi Trip Time Prediction (II) Competition

Inspiration

Added this dataset because competition datasets do not appear in the dataset search and this dataset could help learn basic methods in the area of geo-spatial analysis and trajectory handling

Though GPS-based human trajectory data have been commonly used in travel surveys and human mobility studies, missing data or data gaps that are intrinsically relevant to research reliability remain a critical and challenging issue. This study proposes a novel framework for imputing data gaps based on frequent-pattern mining and time geography, which allows for considering spatio-temporal travel restrictions during imputation by evaluating the spatio-temporal topology relations between the space-time prisms of gaps and corresponding frequent activities or trips. For the validation, the proposed framework is applied to raw GPS trajectories that were collected from 139 participants in Switzerland. In the case study, the temporal and spatio-temporal gaps are artificially generated by randomly choosing activities and trips from the trajectory data. Through comparing the mobility indicators (i.e. duration and distance) calculated from raw data, imputed data, and data with gaps, we quantitatively evaluate the performance of the proposed method in terms of Pearson correlation coefficients and deviation. We further compare the framework with the shortest path interpolation method based on the generated spatio-temporal gaps. The comparison results demonstrate the performance and advantage of the proposed method in imputing gaps from GPS-based human movement data.

This dataset consists of weekly trajectory information of Gulf Stream Warm Core Rings from 2000-2010. This work builds upon Silver et al. (2022a) ( https://doi.org/10.5281/zenodo.6436380) which contained Warm Core Ring trajectory information from 2011 to 2020. Combining the two datasets a total of 21 years of weekly Warm Core Ring trajectories can be obtained. An example of how to use such a dataset can be found in Silver et al. (2022b).

The format of the dataset is similar to that of Silver et al. (2022a), and the following description is adapted from their dataset. This dataset is comprised of individual files containing each ring’s weekly center location and its area for 374 WCRs present between January 1, 2000 and December 31, 2010. Each Warm Core Ring is identified by a unique alphanumeric code 'WEyyyymmddA', where 'WE' represents a Warm Eddy (as identified in the analysis charts); 'yyyymmdd' is the year, month and day of formation; and the last character 'A' represents the sequential sighting of the eddies in a particular year. Continuity of a ring which passes from one year to the next is maintained by the same character in the first sighting. For example, the first ring in 2002 having a trailing alphabet of 'F' indicates that five rings were carried over from 2001 which were still observed on January 1, 2002. Each ring has its own netCDF (.nc) filename following its alphanumeric code. Each file contains 4 variables, “Lon”- the ring center’s weekly longitude, “Lat”- the ring center’s weekly latitude, “Area” - the rings weekly size in km², and “Date” in days - representing the days since Jan 01, 0000.

The process of creating the WCR tracking dataset follows the same methodology of the previously generated WCR census (Gangopadhyay et al., 2019, 2020). The Jenifer Clark’s Gulf Stream Charts used to create this dataset are 2-3 times a week from 2000-2010. Thus, we used approximately 1560 Charts for the 10 years of analysis. All of these charts were reanalyzed between 75° and 55°W using QGIS 2.18.16 (2016) and geo-referenced on a WGS84 coordinate system (Decker, 1986).

Silver, A., Gangopadhyay, A, & Gawarkiewicz, G. (2022a). Warm Core Ring Trajectories in the Northwest Atlantic Slope Sea (2011-2020) (1.0.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.6436380

Silver, A., Gangopadhyay, A., Gawarkiewicz, G., Andres, M., Flierl, G., & Clark, J. (2022b). Spatial Variability of Movement, Structure, and Formation of Warm Core Rings in the Northwest Atlantic Slope Sea. Journal of Geophysical Research: Oceans, 127(8), e2022JC018737. https://doi.org/10.1029/2022JC018737

Gangopadhyay, A., G. Gawarkiewicz, N. Etige, M. Monim and J. Clark, 2019. An Observed Regime Shift in the Formation of Warm Core Rings from the Gulf Stream, Nature - Scientific Reports, https://doi.org/10.1038/s41598-019-48661-9. www.nature.com/articles/s41598-019-48661-9.

Gangopadhyay, A., N. Etige, G. Gawarkiewicz, A. M. Silver, M. Monim and J. Clark, 2020. A Census of the Warm Core Rings of the Gulf Stream (1980-2017). Journal of Geophysical Research, Oceans, 125, e2019JC016033. https://doi.org/10.1029/2019JC016033.

QGIS Development Team. QGIS Geographic Information System (2016).

Decker, B. L. World Geodetic System 1984. World geodetic system 1984 (1986).

This dataset contains the ground-truth intersection regulators for a majority of intersections of the city of Hannover, Germany. The ground-truth information is used in order to apply machine learning techniques on (car) GPS trajectory data in order to automatically detect the intersection regulation.

https://data.uni-hannover.de/dataset/1123552a-7946-4924-bbbc-aa7fbc6a800f/resource/0d5185cf-1a67-4374-97c7-397b65dad394/download/hannover_rules_1.png" alt="Rules">

The GPS trajectories related to specifically this dataset are (also) available under: https://doi.org/10.25835/9bidqxvl

Data Acquisition

The ground-truth information are acquired by visiting them on-site and apply manual labeling of each intersection arm individually. Furthermore, satellite images and street-level images were considered but only on a minor degree as on-site labeling is found to be more precise and up-to-date.

Related Publications:

• Zourlidou, S., Sester, M. and Hu, S. (2022): Recognition of Intersection Traffic Regulations From Crowdsourced Data. Preprints 2022, 2022070012. DOI: https://doi.org/10.20944/preprints202207.0012.v1

• Zourlidou, S., Golze, J. and Sester, M. (2022): Traffic Regulation Recognition using Crowd-Sensed GPS and Map Data: a Hybrid Approach, AGILE GIScience Ser., 3, 22, 2022. https://doi.org/10.5194/agile-giss-3-22-2022

• Cheng, H., Lei, H., Zourlidou, S., Sester, M. (2022): Traffic Control Recognition with an Attention Mechanism Using Speed-Profile and Satellite Imagery data. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences XLIII-B4-2022, S. 287–29. https://doi.org/10.5194/isprs-archives-XLIII-B4-2022-287-2022

• Wang, C., Zourlidou, S., Golze, J. and Sester, M. (2020): Trajectory analysis at intersections for traffic rule identification. Geo-spatial Information Science, 11(4):1-10. https://doi.org/10.1080/10095020.2020.1843374

• Cheng, H., Zourlidou, S. and Sester, M. (2020): Traffic Control Recognition with Speed-Profiles: A Deep Learning Approach. ISPRS Int. J. Geo-Inf. 2020, 9, 652. https://doi.org/10.3390/ijgi9110652

• Golze, J., Zourlidou, S. and Sester, M. (2020): Traffic Regulator Detection Using GPS Trajectories. KN J. Cartogr. Geogr. Inf. https://doi.org/10.1007/s42489-020-00048-x

• Zourlidou, S., Fischer, C. and Sester, M. (2019): Classification of street junctions according to traffic regulators. In: Kyriakidis, P., Hadjimitsis, D., Skarlatos, D. and Mansourian, A., (eds) 2019. Accepted short papers and posters from the 22nd AGILE conference on geo-information science. Cyprus University of Technology 17–20 June 2019, Limassol, Cyprus.

Related Datasets:

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: GPS Trajectory Dataset of the Region of Hannover, Germany. https://doi.org/10.25835/9bidqxvl

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: Traffic Regulator Ground-truth Information for the Chicago Trajectory Dataset. https://doi.org/10.25835/0vifyzqi

• Zourlidou, S., Golze, J. and Sester, M. (2022). Dataset: GPS Trajectory Dataset and Traffic Regulation Information of the Region of Edessa, Greece. https://doi.org/10.25835/v0mzwob3

• Zourlidou, S., Golze, J. and Sester, M. (2020). Dataset: Speed profiles and GPS Trajectories for Traffic Rule Recognition (6 Junctions, Hannover, Germany). https://doi.org/10.25835/0043786

Partical taxi trajectory data in Beijing

We sampled the ground images from the picked videos with a sampling rate of 1/frame/10m to have some visual changes between the consecutive frames, but at the same time, the frames still look relatable to one another. The IMU/GPS data is captured at 1 sample/s. The distance moved in one second varies; the speed of the vehicles is not constant. In this dataset, we care about the visual changes, not the passage of time. So when the distance between every two consecutive locations is greater than 10 m, we resample the video. For more details check "Leveraging Cross-View Geo-Localization With Ensemble Learning And Temporal Awareness".

This dataset contains weekly trajectory information of Gulf Stream Warm Core Rings from 2011-2020. This dataset is comprised of individual files containing each ring’s weekly center location and its area for 282 WCRs present between January 1, 2011 and December 31, 2020. Each Warm Core Ring and is identified by a unique alphanumeric code 'WEyyyymmddA', where 'WE' represents a Warm Eddy (as identified in the analysis charts); 'yyyymmdd' is the year, month and day of formation; and the last character 'A' represents the sequential sighting of the eddies in a particular year. Continuity of a ring which passes from one year to the next is maintained by the same character in the first sighting. For example, the first ring in 2017 having a trailing alphabet of 'E' indicates that four rings were carried over from 2016 which were still observed on January 1, 2017. Each ring has its own netCDF (.nc) filename following its alphanumeric code. Each file contains 4 variables, “Lon”- the ring center’s weekly longitude, “Lat”- the ring center’s weekly latitude, “Area” - the rings weekly size in km², and “Date” in days - representing the week since Jan 01, 0000.

The process of creating the WCR tracking dataset follows the same methodology of the previously generated WCR census (Gangopadhyay et al., 2019, 2020). The Jenifer Clark’s Gulf Stream Charts used to create this dataset are 2-3 times a week from 2011-2020. Thus, we used approximately 1560 Charts for the 10 years of analysis. All of these charts were reanalyzed between 75° and 55°W using QGIS 2.18.16 (2016) and geo-referenced on a WGS84 coordinate system (Decker, 1986).

Gangopadhyay, A., G. Gawarkiewicz, N. Etige, M. Monim and J. Clark, 2019. An Observed Regime Shift in the Formation of Warm Core Rings from the Gulf Stream, Nature - Scientific Reports, https://doi.org/10.1038/s41598-019-48661-9. www.nature.com/articles/s41598-019-48661-9.

Gangopadhyay, A., N. Etige, G. Gawarkiewicz, A. M. Silver, M. Monim and J. Clark, 2020. A Census of the Warm Core Rings of the Gulf Stream (1980-2017). Journal of Geophysical Research, Oceans, 125, e2019JC016033. https://doi.org/10.1029/2019JC016033.

QGIS Development Team. QGIS Geographic Information System (2016).

Decker, B. L. World Geodetic System 1984. World geodetic system 1984 (1986).

The size of the IP Geo-Location Service market was valued at USD XXX million in 2024 and is projected to reach USD XXX million by 2033, with an expected CAGR of XX % during the forecast period.