Published June 4, 2024 | Version 0.1
Dataset Open

Pedestrian street crossing in difficult weather conditions (Pedestrian-DVS)

  • 1. AGH University of Krakow
  • 2. ACC Cyfronet AGH
  • 3. Sano Centre for Computational Medicie

Description

Pedestrian-DVS is a dataset of video frames showing pedestrians crossing the street obtained from CARLA simulation environment.

The dataset consists of two subsets, showing scenes in good and bad weather conditions. The good weather subset consists of 117 videos, while bad weather subset contains 81 videos. Each video consists of 900 frames.

Standard RGB frames are paired with corresponding frames extracted from recordings of dynamic vision sensor. Each frame is labeled using the following format: <frame_number-label.format>. A frame is labeled as positive if, in a given frame, the pedestrian is crossing the street.

For more information, please visit the repository with code used to analyze this dataset or contact the authors.

Contact email:

szmazurek@agh.edu.pl

Files

dvs_datasets.zip

Files (10.9 GB)

Name Size Download all
md5:2b74bca36279650001dd8e929a48a160
10.9 GB Preview Download

Additional details

Related works

Is published in
Preprint: arXiv:2406.00473 (arXiv)

Software

Repository URL
https://github.com/szmazurek/snn_dvs.git
Programming language
Python
Development Status
Wip