uB-VisioGeoloc (2023)

[1] : Imagerie et Vision Artificielle (UR 7535) (Université de Bourgogne)
[2] : Laboratoire d'étude de l'apprentissage et du développement (Université de Bourgogne)
Description :
This dataset contains sequences of pedestrian navigation in an urban environment carried out in the city of Dijon, France, in a variety of representative environments.
Synchronised colour and depth images, as well as GPS and inertial positioning data, are recorded. The colour images are annotated to reference and locate elements of significance to pedestrians (cars, pavements, signs, etc.).
This dataset was produced as part of a project to develop a mobility assistance system for the visually impaired using sound substitution. For each sequence, a sound file corresponding to a substitution method is proposed.
Disciplines :
computer science, artificial intelligence (engineering science), computer science, software engineering (engineering science), engineering, industrial (engineering science)

General metadata

Data acquisition date : from 28 Mar 2023 ongoing
Data acquisition methods :
  • Observational data :
    To record this data, a person equipped with a recording system travelled around the city of Dijon. The system consisted of an Intel Realsense D435 RGB-D camera (rvb-d), an Adafruit BNO055 Inertial Measurement Unit (IMU-100Hz) sensor and a GPS antenna (10Hz). Each sensor had a dedicated thread operating independently of the others to ensure synchronisation between the rate of data acquisition and the recording process.
    A synthetic sequence modelling the Place Darcy in Dijon (produced from a LIDAR scan) was added to enrich the dataset.
    The image-sound conversion generates spatialised sounds whose frequency depends on elevation (from 250 Hz to 1492 Hz), which is convolved with the HRTFs in the CIPIC database to obtain a stereophonic sound spatialised in azimuth and elevation.
Update periodicity : as needed
Language : English (eng)
Formats : audio/mpeg, image/png, text/xml
Audience : Research, Informal Education


Spatial coverage :

  • Dijon: latitude between 47° 22' 39" N and 47° 17' 10" N, longitude between 4° 57' 44" E and 5° 6' 7" E
Publications :
  • An image sequences dataset of pedestrian navigation including geolocalised-inertial information and spatial sound rendering of the urban environment's obstacles (doi:10.1016/j.dib.2024.110088)
Projects and funders :
Additional information :
Data collected as part of the theses of Florian Scalvini (under the supervision of Professor Julien Dubois (ImVia), Professor Maxime Ambard (LEAD) and Professor Cyrille Migniot (ImVia)) and Camille Bordeau (under the supervision of Professor Emmanuel Bigand (LEAD) and Professor Maxime Ambard (LEAD)).
Record created 13 Jul 2023 by Cyrille Migniot.
Last modification : 3 Apr 2024.
Local identifier: FR-13002091000019-2023-07-13.


dat@uB is a sub-portal of dat@UBFC, a metadata catalogue for research data produced at UBFC.

Terms of use
Université de Bourgogne, Université de Franche-Comté, UTBM, AgroSup Dijon, ENSMM, BSB, Arts des Metiers