Multi-Scene Infrastructure and Vehicle Data for Advancing Collective Perception in Autonomous Driving
The CoopScenes dataset is a large-scale, multi-scene dataset designed to support research in collective perception,
real-time sensor registration, and cooperative intelligent systems for urban mobility. The dataset features synchronized
multi-sensor data from both an ego-vehicle and infrastructure sensors, providing researchers with high-quality data for
machine learning and sensor fusion applications.
✔ 104 minutes of synchronized data at 10 Hz, totaling 62,000 frames
✔ Highly accurate synchronization with a mean deviation of 2.3 ms
✔ Precise point cloud registration between the ego-vehicle and infrastructure sensors
✔ Automated annotation pipelines for object labeling
✔ Open-source anonymization for faces and license plates with BlurScene
✔ Diverse scenarios: public transport hubs, construction sites, and high-speed roads across three cities in Stuttgart, Germany
✔ Total dataset size: 527 GB in .4mse
format, accessible via our development kit
The dataset can be accessed via official CoopScenes website and used with our development kit.
📢 INFO: The data will be fully published upon the official publication announcement.
To use the dataset, simply install our provided PyPi package:
python3 -m pip install CoopScenes
git clone https://github.com/MarcelVSHNS/CoopScenes.git
cd CoopScenes
python -m venv venv # install with apt-get install python3-venv
source venv/bin/activate
pip install -r requirements.txt
You can find detailed examples in the Colab notebook.
import coopscenes as cs
sample_record = cs.DataRecord("/content/example_record_1.4mse")
frame = sample_record[0]
frame.vehicle.cameras.STEREO_LEFT.show() # PIL Image
@misc{vosshans2024aeifdatacollectiondataset,
author = {Marcel Vosshans and Alexander Baumann and Matthias Drueppel and Omar Ait-Aider and Ralf Woerner and Youcef Mezouar and Thao Dang and Markus Enzweiler},
title = {The AEIF Data Collection: A Dataset for Infrastructure-Supported Perception Research with Focus on Public Transportation},
url = {https://arxiv.org/abs/2407.08261},
year = {2024},
}
This dataset is released under the MIT License.
Enjoy using CoopScenes! 🚀