Reflectance Adaptive Filtering Improves Intrinsic Image Estimation
Thomas Nestmeyer, Peter V. Gehler

TL;DR
This paper demonstrates that simple pixel-wise decisions and explicit prior knowledge through bilateral filtering can achieve state-of-the-art results in intrinsic image decomposition, challenging the perceived dominance of learning-based methods.
Contribution
It introduces a bilateral filtering approach that incorporates strong prior knowledge, improving intrinsic image estimation and establishing a new state-of-the-art benchmark on IIW.
Findings
A standard signal processing technique matches state-of-the-art performance.
A CNN loss function for dense reflectance prediction provides a strong baseline.
Explicit priors via bilateral filtering enhance intrinsic image decomposition results.
Abstract
Separating an image into reflectance and shading layers poses a challenge for learning approaches because no large corpus of precise and realistic ground truth decompositions exists. The Intrinsic Images in the Wild~(IIW) dataset provides a sparse set of relative human reflectance judgments, which serves as a standard benchmark for intrinsic images. A number of methods use IIW to learn statistical dependencies between the images and their reflectance layer. Although learning plays an important role for high performance, we show that a standard signal processing technique achieves performance on par with current state-of-the-art. We propose a loss function for CNN learning of dense reflectance predictions. Our results show a simple pixel-wise decision, without any context or prior knowledge, is sufficient to provide a strong baseline on IIW. This sets a competitive baseline which only…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Code & Models
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsImage Enhancement Techniques · Advanced Image Fusion Techniques · Color Science and Applications
