Zero-shot Inexact CAD Model Alignment from a Single Image
This enables 3D scene reconstruction from images for a broad range of object categories without costly supervised data, though it is incremental over prior weakly supervised methods.
The paper tackles the problem of aligning inexact 3D CAD models to objects in single images without pose annotations, achieving state-of-the-art results with a +4.3% mean alignment accuracy improvement on ScanNet25k and generalization to 20 novel categories on SUN2CAD.
One practical approach to infer 3D scene structure from a single image is to retrieve a closely matching 3D model from a database and align it with the object in the image. Existing methods rely on supervised training with images and pose annotations, which limits them to a narrow set of object categories. To address this, we propose a weakly supervised 9-DoF alignment method for inexact 3D models that requires no pose annotations and generalizes to unseen categories. Our approach derives a novel feature space based on foundation features that ensure multi-view consistency and overcome symmetry ambiguities inherent in foundation features using a self-supervised triplet loss. Additionally, we introduce a texture-invariant pose refinement technique that performs dense alignment in normalized object coordinates, estimated through the enhanced feature space. We conduct extensive evaluations on the real-world ScanNet25k dataset, where our method outperforms SOTA weakly supervised baselines by +4.3% mean alignment accuracy and is the only weakly supervised approach to surpass the supervised ROCA by +2.7%. To assess generalization, we introduce SUN2CAD, a real-world test set with 20 novel object categories, where our method achieves SOTA results without prior training on them.