A 2.7K-parameter physically-informed network that simulates ColorChecker images under a given illumination to derive illumination-adaptive raw-to-raw color transforms between cameras, trainable with or without paired data.
Title resolution pending
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
-
Beyond Calibration: Physically Informed Learning for Raw-to-Raw Mapping
A 2.7K-parameter physically-informed network that simulates ColorChecker images under a given illumination to derive illumination-adaptive raw-to-raw color transforms between cameras, trainable with or without paired data.