Learning multiplane images from single views with self-supervision

TL;DR

This paper introduces CycleMPI, a self-supervised framework that learns multiplane image representations from single images, enabling high-quality novel view synthesis without stereo data, and generalizes well to challenging scenarios.

Contribution

The paper presents a novel cyclic training strategy for learning multiplane images from single views without stereo data, improving generalization in view synthesis tasks.

Findings

Achieves state-of-the-art results in zero-shot view synthesis on stereo datasets.

Performs well on challenging datasets like RealEstate10K and Mannequin Challenge.

Produces high-quality qualitative results on Places II.

Abstract

Generating static novel views from an already captured image is a hard task in computer vision and graphics, in particular when the single input image has dynamic parts such as persons or moving objects. In this paper, we tackle this problem by proposing a new framework, called CycleMPI, that is capable of learning a multiplane image representation from single images through a cyclic training strategy for self-supervision. Our framework does not require stereo data for training, therefore it can be trained with massive visual data from the Internet, resulting in a better generalization capability even for very challenging cases. Although our method does not require stereo data for supervision, it reaches results on stereo datasets comparable to the state of the art in a zero-shot scenario. We evaluated our method on RealEstate10K and Mannequin Challenge datasets for view synthesis and presented qualitative results on Places II dataset.