An Empirical Study on Multi-Domain Robust Semantic Segmentation

TL;DR

This paper investigates training a single semantic segmentation model across multiple domains, analyzing factors like data augmentation, training strategies, and model capacity to improve generalization, and proposes a robust solution that ranks second in a competitive benchmark.

Contribution

It provides a detailed analysis of multi-domain training challenges and introduces a robust method that enhances generalization across diverse datasets.

Findings

The proposed method improves cross-domain performance.

It ranks 2nd on RVC 2022 semantic segmentation task.

The approach works with only one-third of the dataset size of the top model.

Abstract

How to effectively leverage the plentiful existing datasets to train a robust and high-performance model is of great significance for many practical applications. However, a model trained on a naive merge of different datasets tends to obtain poor performance due to annotation conflicts and domain divergence.In this paper, we attempt to train a unified model that is expected to perform well across domains on several popularity segmentation datasets.We conduct a detailed analysis of the impact on model generalization from three aspects of data augmentation, training strategies, and model capacity.Based on the analysis, we propose a robust solution that is able to improve model generalization across domains.Our solution ranks 2nd on RVC 2022 semantic segmentation task, with a dataset only 1/3 size of the 1st model used.