Polarimetric image augmentation

TL;DR

This paper introduces a novel regularized augmentation method for polarimetric images to improve scene segmentation in urban robotics, addressing the challenge of limited data and the unique properties of polarization data.

Contribution

The paper proposes a new augmentation technique tailored for polarimetric images, enhancing deep learning segmentation performance under challenging conditions.

Findings

18.1% average IoU improvement with regularized augmentation

Effective scene segmentation in urban environments using polarization data

Addresses data scarcity in polarimetric deep learning applications

Abstract

Robotics applications in urban environments are subject to obstacles that exhibit specular reflections hampering autonomous navigation. On the other hand, these reflections are highly polarized and this extra information can successfully be used to segment the specular areas. In nature, polarized light is obtained by reflection or scattering. Deep Convolutional Neural Networks (DCNNs) have shown excellent segmentation results, but require a significant amount of data to achieve best performances. The lack of data is usually overcomed by using augmentation methods. However, unlike RGB images, polarization images are not only scalar (intensity) images and standard augmentation techniques cannot be applied straightforwardly. We propose to enhance deep learning models through a regularized augmentation procedure applied to polarimetric data in order to characterize scenes more effectively under challenging conditions. We subsequently observe an average of 18.1% improvement in IoU between non augmented and regularized training procedures on real world data.