Improving Panoptic Segmentation for Nighttime or Low-Illumination Urban Driving Scenes

TL;DR

This paper enhances panoptic segmentation for nighttime urban driving scenes by using domain translation with CycleGAN to generate annotated nighttime images, significantly improving performance and robustness in low-light conditions.

Contribution

It introduces two novel methods leveraging CycleGAN for domain translation to improve panoptic segmentation accuracy and robustness in poor illumination urban scenes.

Findings

Over 10% improvement in PQ on converted Cityscapes dataset

Enhanced robustness to varied nighttime environments

Significant gains in RQ, SQ, mIoU, and AP50 metrics

Abstract

Autonomous vehicles and driving systems use scene parsing as an essential tool to understand the surrounding environment. Panoptic segmentation is a state-of-the-art technique which proves to be pivotal in this use case. Deep learning-based architectures have been utilized for effective and efficient Panoptic Segmentation in recent times. However, when it comes to adverse conditions like dark scenes with poor illumination or nighttime images, existing methods perform poorly in comparison to daytime images. One of the main factors for poor results is the lack of sufficient and accurately annotated nighttime images for urban driving scenes. In this work, we propose two new methods, first to improve the performance, and second to improve the robustness of panoptic segmentation in nighttime or poor illumination urban driving scenes using a domain translation approach. The proposed approach makes use of CycleGAN (Zhu et al., 2017) to translate daytime images with existing panoptic annotations into nighttime images, which are then utilized to retrain a Panoptic segmentation model to improve performance and robustness under poor illumination and nighttime conditions. In our experiments, Approach-1 demonstrates a significant improvement in the Panoptic segmentation performance on the converted Cityscapes dataset with more than +10% PQ, +12% RQ, +2% SQ, +14% mIoU and +10% AP50 absolute gain. Approach-2 demonstrates improved robustness to varied nighttime driving environments. Both the approaches are supported via comprehensive quantitative and qualitative analysis.