A Fine-Grained Dataset and its Efficient Semantic Segmentation for Unstructured Driving Scenarios

TL;DR

This paper introduces TAS500, a detailed dataset for semantic segmentation in unstructured driving environments, emphasizing the importance of fine-grained classes for improved accuracy and efficiency in autonomous driving tasks.

Contribution

The paper presents TAS500, a new dataset with fine-grained classes for unstructured outdoor scenes, and evaluates model performance emphasizing efficiency and boundary accuracy.

Findings

Fine-grained classes improve segmentation accuracy.

Models perform better on class boundaries with TAS500.

Dataset and pretrained models are publicly available.

Abstract

Research in autonomous driving for unstructured environments suffers from a lack of semantically labeled datasets compared to its urban counterpart. Urban and unstructured outdoor environments are challenging due to the varying lighting and weather conditions during a day and across seasons. In this paper, we introduce TAS500, a novel semantic segmentation dataset for autonomous driving in unstructured environments. TAS500 offers fine-grained vegetation and terrain classes to learn drivable surfaces and natural obstacles in outdoor scenes effectively. We evaluate the performance of modern semantic segmentation models with an additional focus on their efficiency. Our experiments demonstrate the advantages of fine-grained semantic classes to improve the overall prediction accuracy, especially along the class boundaries. The dataset and pretrained model are available at mucar3.de/icpr2020-tas500.