OASIS adds a lightweight structure-refinement decoder and evidential-learning loss to a memory-based video object segmentation backbone, yielding modest but consistent benchmark improvements over Cutie.
Evidential deep learning to quantify classification uncertainty,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Structure Matters: Revisiting Boundary Refinement in Video Object Segmentation
OASIS adds a lightweight structure-refinement decoder and evidential-learning loss to a memory-based video object segmentation backbone, yielding modest but consistent benchmark improvements over Cutie.