DCGNet combines dynamic multi-granularity detection, underwater physics priors for degradation restoration, spatial Gaussian saliency, and a diffusion transformer to outperform prior methods on underwater SOD benchmarks.
Learning heavily- degraded prior for underwater object detection,
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
-
Rethinking Conditional Generation for Underwater Salient Object Detection
DCGNet combines dynamic multi-granularity detection, underwater physics priors for degradation restoration, spatial Gaussian saliency, and a diffusion transformer to outperform prior methods on underwater SOD benchmarks.