Geometric Correspondence Fields: Learned Differentiable Rendering for 3D Pose Refinement in the Wild
This addresses the problem of precise 3D object alignment in real-world images for computer vision applications, representing a novel method rather than an incremental improvement.
The paper tackles 3D pose refinement for objects in the wild by introducing a differentiable rendering approach that compares images in an optimized feature space and uses learned geometric correspondence fields, achieving up to 55% relative improvement over state-of-the-art methods on the Pix3D dataset.
We present a novel 3D pose refinement approach based on differentiable rendering for objects of arbitrary categories in the wild. In contrast to previous methods, we make two main contributions: First, instead of comparing real-world images and synthetic renderings in the RGB or mask space, we compare them in a feature space optimized for 3D pose refinement. Second, we introduce a novel differentiable renderer that learns to approximate the rasterization backward pass from data instead of relying on a hand-crafted algorithm. For this purpose, we predict deep cross-domain correspondences between RGB images and 3D model renderings in the form of what we call geometric correspondence fields. These correspondence fields serve as pixel-level gradients which are analytically propagated backward through the rendering pipeline to perform a gradient-based optimization directly on the 3D pose. In this way, we precisely align 3D models to objects in RGB images which results in significantly improved 3D pose estimates. We evaluate our approach on the challenging Pix3D dataset and achieve up to 55% relative improvement compared to state-of-the-art refinement methods in multiple metrics.