MIST: Multi-Domain Synthetic Dataset for Rural Autonomous Driving

MIST is a large-scale multi-domain synthetic dataset designed for rural driving scenarios. It provides 32 balanced domain configurations structured by season, time of day, and weather. The dataset was created by jongwonryu and is hosted on Hugging Face, with a last recorded update in February 2026.

Use Cases

• Training perception models for rural driving based on the described multi-domain synthetic imagery.
• Benchmarking domain adaptation algorithms across the 32 structured domain configurations.
• Simulating and testing autonomous vehicle behavior in varied rural conditions like weather and time of day.

Strengths

• Explicitly structured domain factors forming 32 balanced configurations.
• Large-scale synthetic data generation focused on rural driving scenarios.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and file size are unknown, which may limit suitability assessment.
• The dataset is synthetic, which may not fully capture the complexity of real-world rural driving.

Provenance

Source

huggingface

Collection Method

Synthetic data generated using the simulator slowroads.io.

Freshness

Last updated 2026-02-04 11:24:34; freshness should be verified.

Geography

Rural driving scenarios (non-specific).