Shelf-Supervised Mesh Prediction in the Wild

TL;DR

This paper introduces a shelf-supervised learning approach for 3D shape and pose estimation from single images, using only segmentation supervision from existing recognition systems, enabling scalable multi-category 3D reconstruction.

Contribution

It presents a novel method that infers 3D shape and pose from unstructured image collections with minimal supervision, scalable to 50 categories in the wild.

Findings

Effective 3D shape and pose inference from single images.

Scalable to 50 categories in real-world datasets.

Outperforms existing methods in multi-category 3D reconstruction.

Abstract

We aim to infer 3D shape and pose of object from a single image and propose a learning-based approach that can train from unstructured image collections, supervised by only segmentation outputs from off-the-shelf recognition systems (i.e. 'shelf-supervised'). We first infer a volumetric representation in a canonical frame, along with the camera pose. We enforce the representation geometrically consistent with both appearance and masks, and also that the synthesized novel views are indistinguishable from image collections. The coarse volumetric prediction is then converted to a mesh-based representation, which is further refined in the predicted camera frame. These two steps allow both shape-pose factorization from image collections and per-instance reconstruction in finer details. We examine the method on both synthetic and real-world datasets and demonstrate its scalability on 50 categories in the wild, an order of magnitude more classes than existing works.