MM-UAVBench: A Multimodal Benchmark for Low-Altitude UAV Scenarios

A multimodal benchmark designed to evaluate Multimodal Large Language Models (MLLMs) in low-altitude UAV scenarios. It contains 19 tasks across three capability dimensions: perception, cognition, and planning. The dataset was created by author 'daisq' and was last updated on January 18, —.

Use Cases

• Benchmarking model perception abilities based on the described UAV-specific low-altitude scenarios
• Evaluating cognitive reasoning for UAVs based on the described task dimensions
• Testing autonomous planning capabilities for drones based on the benchmark's design
• Comparing MLLM performance across the 19 described tasks

Strengths

• Designed with 19 specific tasks for structured evaluation
• Focuses on three core capability dimensions: perception, cognition, and planning
• Targets a specific application domain: low-altitude UAV scenarios

Limitations

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

Provenance

Source

huggingface

Collection Method

Likely curated for benchmarking purposes, as described.

Time Range

null

Freshness

Last updated 2026-01 13:16:18; freshness should be verified

Geography

null