Poppy: Polarization-based Plug-and-Play Guidance for Enhancing Monocular Normal Estimation

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Kim, Irene, Chakkera, Sai Tanmay Reddy, Graikos, Alexandros, Samaras, Dimitris, Dave, Akshat
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866911552478642176
author Kim, Irene
Chakkera, Sai Tanmay Reddy
Graikos, Alexandros
Samaras, Dimitris
Dave, Akshat
author_facet Kim, Irene
Chakkera, Sai Tanmay Reddy
Graikos, Alexandros
Samaras, Dimitris
Dave, Akshat
contents Monocular surface normal estimators trained on large-scale RGB-normal data often perform poorly in the edge cases of reflective, textureless, and dark surfaces. Polarization encodes surface orientation independently of texture and albedo, offering a physics-based complement for these cases. Existing polarization methods, however, require multi-view capture or specialized training data, limiting generalization. We introduce Poppy, a training-free framework that refines normals from any frozen RGB backbone using single-shot polarization measurements at test time. Keeping backbone weights frozen, Poppy optimizes per-pixel offsets to the input RGB and output normal along with a learned reflectance decomposition. A differentiable rendering layer converts the refined normals into polarization predictions and penalizes mismatches with the observed signal. Across seven benchmarks and three backbone architectures (diffusion, flow, and feed-forward), Poppy reduces mean angular error by 23-26% on synthetic data and 6-16% on real data. These results show that guiding learned RGB-based normal estimators with polarization cues at test time refines normals on challenging surfaces without retraining.
format Preprint
id arxiv_https___arxiv_org_abs_2603_27891
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Poppy: Polarization-based Plug-and-Play Guidance for Enhancing Monocular Normal Estimation
Kim, Irene
Chakkera, Sai Tanmay Reddy
Graikos, Alexandros
Samaras, Dimitris
Dave, Akshat
Computer Vision and Pattern Recognition
Monocular surface normal estimators trained on large-scale RGB-normal data often perform poorly in the edge cases of reflective, textureless, and dark surfaces. Polarization encodes surface orientation independently of texture and albedo, offering a physics-based complement for these cases. Existing polarization methods, however, require multi-view capture or specialized training data, limiting generalization. We introduce Poppy, a training-free framework that refines normals from any frozen RGB backbone using single-shot polarization measurements at test time. Keeping backbone weights frozen, Poppy optimizes per-pixel offsets to the input RGB and output normal along with a learned reflectance decomposition. A differentiable rendering layer converts the refined normals into polarization predictions and penalizes mismatches with the observed signal. Across seven benchmarks and three backbone architectures (diffusion, flow, and feed-forward), Poppy reduces mean angular error by 23-26% on synthetic data and 6-16% on real data. These results show that guiding learned RGB-based normal estimators with polarization cues at test time refines normals on challenging surfaces without retraining.
title Poppy: Polarization-based Plug-and-Play Guidance for Enhancing Monocular Normal Estimation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.27891