A label-free metric-guided fusion of complementary features from visual foundation models yields consistent gains in dense prediction tasks with improved object semantics and boundary localization.
Quantifying the limits of segmentation foun- dation models: Modeling challenges in segmenting tree-like and low-contrast objects.arXiv preprint arXiv:2412.04243
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CV 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
GLASSNet outperforms prior methods on salient object detection benchmarks by freezing SAMv2, adding a spatially aware adapter, and fusing outputs from global and local decoders.
citing papers explorer
-
Metric-Guided Feature Fusion of Visual Foundation Models for Segmentation Tasks
A label-free metric-guided fusion of complementary features from visual foundation models yields consistent gains in dense prediction tasks with improved object semantics and boundary localization.
-
Global-Local Feature Decoding with Adapter-Guided SAMv2 for Salient Object Detection
GLASSNet outperforms prior methods on salient object detection benchmarks by freezing SAMv2, adding a spatially aware adapter, and fusing outputs from global and local decoders.