UAV-OVO benchmark exposes large ID/OOD performance gaps in video action recognition due to low-to-high depression viewpoint shifts, and LATER uses LoRA subspace anchoring for test-time feature re-centering to reduce drift.
de Melo, Stephen M
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
UAV-OVO: Out-of-Viewpoint Generalization in UAV Action Recognition
UAV-OVO benchmark exposes large ID/OOD performance gaps in video action recognition due to low-to-high depression viewpoint shifts, and LATER uses LoRA subspace anchoring for test-time feature re-centering to reduce drift.