UAVIT-1M: Instruction Tuning Dataset for Low-Altitude UAV Visual Understanding

UAVIT-1M is a large instruction tuning dataset designed to enhance visual understanding capabilities for low-altitude unmanned aerial vehicles (UAVs). It supports 11 image-level and region-level tasks. The dataset was created by ZhanYang-nwpu and released on Hugging Face in May 2025.

Use Cases

• Training vision-language models for UAV scene interpretation based on the described instruction tuning format.
• Benchmarking model performance on low-altitude aerial tasks based on the 11 supported task types.
• Fine-tuning models for region-level analysis in drone imagery as mentioned in the description.

Strengths

• Designed for 11 distinct visual understanding tasks for UAVs.
• Reported as the first largest and most comprehensive dataset of its kind for low-altitude UAVs.
• Released with an associated benchmark (UAVBench) for evaluation.

Limitations

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

Provenance

Source

ZhanYang-nwpu via Hugging Face.

Collection Method

Likely collected and annotated from low-altitude UAV imagery, but specific methodology is not detailed.

Time Range

null

Freshness

Last updated 2025-05-16 12:36:49.

Geography

null