Aerial Drone Photography& Videography
EPFLLUTS

Downloads

DOWNLOAD DATA

Do not forget to visit the πNEUMA Blog categorised by code to download the openly shared codes from various open source projects

pNEUMA dataset © 2020 by E. Barmpounakis and N. Geroliminis is licensed under CC BY-NC 4.0

FAQ

pNEUMA dataset © 2020 by E. Barmpounakis and N. Geroliminis is licensed under CC BY-NC 4.0. Thus, as also described in the terms and conditions, the specific dataset is distributed for free and can be used only for non-commercial purposesFinally, for any publication that utilises the pNEUMA dataset, authors should include a reference to this article and include “Data source: pNEUMA – open-traffic.epfl.ch” in the acknowledgment section.
For each .csv file the following apply:
– each row represents the data of a single vehicle
– the first 10 columns in the 1st row include the columns’ names
– the first 4 columns include information about the trajectory like the unique trackID, the type of vehicle, the distance traveled in meters and the average speed of the vehicle in km/h
– the last 6 columns are then repeated every 6 columns based on the time frequency. For example, column_5 contains the latitude of the vehicle at time column_10, and column­­­_11 contains the latitude of the vehicle at time column_16.
– Speed is in km/h, Longitudinal and Lateral Acceleration in m/sec2 and time in seconds.

There are 6 types of vehicles. These are Car, Taxi, Bus, Medium Vehicle, Heavy Vehicle, Motorcycle.

The dimensions of the different types of vehicles are given below in meters:

  • Car and Taxi:  5 x 2
  • Medium Vehicle: 5.83 x 2.67
  • Heavy Vehicle: 12.5 x 3.33
  • Bus: 12.5 x 4
  • Motorcycle: 2.5 x 1

There are 58 bus stations, most of them being curbside stops.

A complete methodology to extract lane-wise information is described in Barmpounakis, E., Sauvin, G. M., & Geroliminis, N. (2020). Lane Detection and Lane-Changing Identification with High-Resolution Data from a Swarm of Drones. Transportation Research Record, https://doi.org/10.1177/0361198120920627 (click here).

One of the drones instead of 25 FPS recorded the traffic streams in 23.97 FPS. Since the frequency of the explored dataset is depended on the number of frames, this affects the time frequency of the dataset. We advise to use resampling techniques for consistency with the rest of the datasets.

These trajectories are mostly conducted by motorcycles that sometimes may have unconventional trajectories, for example move on pedestrian streets.

Another reason can be drone movements during that reduced the quality of the georegistration process. Such an example is dataset 20181024_d4_0830_0900.csv

For various technical issues one or more drones may have stopped recording. Thus, the corresponding drone’s data will be missed, while all the other drones would still keep working.

The information about the signal timing plans for the signalised intersections is not available. However, since in Athens they are pre-timed, this information can be obtained by the dataset (recognising cycle lengths, splits and offsets etc.).

   
    
downloads page
    
downloads page