AI RESEARCH
RPBA-Net: An Interpretable Residual Pyramid Bilateral Affine Network for RAW-Domain ISP Enhancement
arXiv CS.CV
•
ArXi:2605.03626v1 Announce Type: new To address module fragmentation, uninterpretable mappings, and deployment constraints in RAW-domain saicing, color correction, and detail enhancement, this paper proposes RPBA-Net, an interpretable residual pyramid bilateral affine network for RAW-domain ISP enhancement. Given packed RAW as input, the method performs residual affine base reconstruction by estimating a base RGB representation and learning identity-guided residual affine corrections, thereby unifying saicing and enhancement.