HS3-Bench: A Benchmark and Strong Baseline for Hyperspectral Semantic Segmentation in Driving Scenarios

TL;DR

This paper introduces HS3-Bench, a comprehensive benchmark for hyperspectral semantic segmentation in driving scenarios, providing standardized evaluation tools and establishing strong baseline models that outperform previous methods.

Contribution

It creates the first standardized benchmark for hyperspectral semantic segmentation in driving, combining datasets, metrics, and protocols, and offers strong baseline models that surpass prior state-of-the-art.

Findings

Baseline models outperform previous state-of-the-art results.

Leveraging RGB data benefits hyperspectral segmentation more than additional spectral channels.

The benchmark facilitates systematic evaluation and comparison of hyperspectral segmentation methods.

Abstract

Semantic segmentation is an essential step for many vision applications in order to understand a scene and the objects within. Recent progress in hyperspectral imaging technology enables the application in driving scenarios and the hope is that the devices perceptive abilities provide an advantage over RGB-cameras. Even though some datasets exist, there is no standard benchmark available to systematically measure progress on this task and evaluate the benefit of hyperspectral data. In this paper, we work towards closing this gap by providing the HyperSpectral Semantic Segmentation benchmark (HS3-Bench). It combines annotated hyperspectral images from three driving scenario datasets and provides standardized metrics, implementations, and evaluation protocols. We use the benchmark to derive two strong baseline models that surpass the previous state-of-the-art performances with and without pre-training on the individual datasets. Further, our results indicate that the existing learning-based methods benefit more from leveraging additional RGB training data than from leveraging the additional hyperspectral channels. This poses important questions for future research on hyperspectral imaging for semantic segmentation in driving scenarios. Code to run the benchmark and the strong baseline approaches are available under https://github.com/nickstheisen/hyperseg.