Cross-modal Learning for Image-Guided Point Cloud Shape Completion
This addresses the problem of incomplete 3D data reconstruction for applications like robotics or computer vision, but it is incremental as it builds on existing multimodal learning techniques.
The paper tackles point cloud shape completion guided by an auxiliary image, achieving significant improvements over state-of-the-art supervised methods in both unimodal and multimodal settings, with the weakly-supervised approach outperforming many supervised methods and being competitive with the latest supervised models.
In this paper we explore the recent topic of point cloud completion, guided by an auxiliary image. We show how it is possible to effectively combine the information from the two modalities in a localized latent space, thus avoiding the need for complex point cloud reconstruction methods from single views used by the state-of-the-art. We also investigate a novel weakly-supervised setting where the auxiliary image provides a supervisory signal to the training process by using a differentiable renderer on the completed point cloud to measure fidelity in the image space. Experiments show significant improvements over state-of-the-art supervised methods for both unimodal and multimodal completion. We also show the effectiveness of the weakly-supervised approach which outperforms a number of supervised methods and is competitive with the latest supervised models only exploiting point cloud information.