Raw2Event: Paired Raw, RGB, and Event Camera Data for CIFAR-10 Classes

Raw2Event is a paired dataset and benchmark containing Bayer raw, ISP-RGB, and real-event camera data. The data was captured by displaying CIFAR-10 image samples on a screen while a robotic arm with a Raspberry Pi camera and a DAVIS346 event camera followed a fixed circular trajectory. The dataset was created by the author 'raw2event' and was last updated on the Hugging Face platform in May 2026.

Use Cases

• Benchmarking frame-to-event simulation algorithms based on the paired raw/RGB/event data.
• Training models for event-based vision tasks using the synchronized real-event camera recordings.
• Developing image processing pipelines that leverage the paired Bayer raw and ISP-RGB video streams.
• Researching robotic perception systems using data captured from a moving robotic arm platform.

Strengths

• Provides synchronized data from three modalities: Bayer raw, ISP-RGB, and real-event camera.
• Built on the established CIFAR-10 image classes, providing a known semantic foundation.
• Data was captured using a robotic arm following a fixed circular trajectory, suggesting controlled acquisition.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and dataset size are unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

huggingface

Collection Method

Captured by displaying CIFAR-10 samples on a screen and recording with a Raspberry Pi Camera Module 3 and a DAVIS346 event camera mounted on a Dobot robotic arm.

Freshness

Last updated 2026-05-24 20:52:50; freshness should be verified.