Factor Unfolding Network for concealed object segmentation
November 20, 2026
The Factor Unfolding Network (FUN) separates semantic content from degradation effects to improve concealed object segmentation. It extends the Retinex basis to explicitly factorize content-degradation formation, allowing the model to use the content factor as the primary source of semantic evidence.
HOW THIS AFFECTS YOU
●
researcherThis method offers a way to improve segmentation in noisy or low-visibility environments by separating signal from noise.