Semantic RGB-D Image Synthesis

TL;DR

This paper introduces a multi-modal semantic RGB-D image synthesis method that generates realistic RGB-D images from semantic labels, improving data diversity and segmentation accuracy in privacy-sensitive applications.

Contribution

It proposes a novel generator and discriminator architecture for multi-modal data that enhances realism and semantic consistency in synthesized images.

Findings

Outperforms previous uni-modal methods significantly

Mixing real and generated images improves segmentation accuracy

Generates realistic RGB-D images from semantic label maps

Abstract

Collecting diverse sets of training images for RGB-D semantic image segmentation is not always possible. In particular, when robots need to operate in privacy-sensitive areas like homes, the collection is often limited to a small set of locations. As a consequence, the annotated images lack diversity in appearance and approaches for RGB-D semantic image segmentation tend to overfit the training data. In this paper, we thus introduce semantic RGB-D image synthesis to address this problem. It requires synthesising a realistic-looking RGB-D image for a given semantic label map. Current approaches, however, are uni-modal and cannot cope with multi-modal data. Indeed, we show that extending uni-modal approaches to multi-modal data does not perform well. In this paper, we therefore propose a generator for multi-modal data that separates modal-independent information of the semantic layout from the modal-dependent information that is needed to generate an RGB and a depth image, respectively. Furthermore, we propose a discriminator that ensures semantic consistency between the label maps and the generated images and perceptual similarity between the real and generated images. Our comprehensive experiments demonstrate that the proposed method outperforms previous uni-modal methods by a large margin and that the accuracy of an approach for RGB-D semantic segmentation can be significantly improved by mixing real and generated images during training.