Rethinking Inductive Biases for Surface Normal Estimation
This addresses the need for accurate surface normal estimation in computer vision, with incremental improvements over existing methods.
The paper tackles the problem of surface normal estimation by proposing a method that uses per-pixel ray direction and encodes relationships between neighboring normals, resulting in improved generalization with a smaller dataset compared to a ViT-based state-of-the-art model.
Despite the growing demand for accurate surface normal estimation models, existing methods use general-purpose dense prediction models, adopting the same inductive biases as other tasks. In this paper, we discuss the inductive biases needed for surface normal estimation and propose to (1) utilize the per-pixel ray direction and (2) encode the relationship between neighboring surface normals by learning their relative rotation. The proposed method can generate crisp - yet, piecewise smooth - predictions for challenging in-the-wild images of arbitrary resolution and aspect ratio. Compared to a recent ViT-based state-of-the-art model, our method shows a stronger generalization ability, despite being trained on an orders of magnitude smaller dataset. The code is available at https://github.com/baegwangbin/DSINE.