Benchmark RadarUAV-OVO
UAV-OVO is an out-of-viewpoint generalization benchmark for UAV action recognition. It evaluates models on low-depression viewpoint training data and tests on high-depression viewpoint out-of-distribution data, with class-matched ID/OOD splits.
- Released
- 2026-05-25
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
Standard UAV action recognition benchmarks often overlook viewpoint shifts, leading to models that rely on viewpoint-specific shortcuts. UAV-OVO provides a controlled testbed to measure and improve robustness to viewpoint changes, which is critical for real-world UAV deployment.
Motivation
UAV action recognition faces a deployment shift that standard benchmarks often obscure: a model trained on UAV footage captured from low-depression viewpoints may be required to recognize the same action classes from high-depression viewpoints.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.