MSeg: A Composite Dataset for Multi-Domain Semantic Segmentation
Authors:
- 1 June 2020
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 2876-2885
Abstract
We present MSeg, a composite dataset that unifies se- mantic segmentation datasets from different domains. A naive merge of the constituent datasets yields poor performance due to inconsistent taxonomies and annotation practices. We reconcile the taxonomies and bring the pixel-level annotations into alignment by relabeling more than 220,000 object masks in more than 80,000 images. The resulting composite dataset enables training a single semantic segmentation model that functions effectively across domains and generalizes to datasets that were not seen during training. We adopt zero-shot cross-dataset transfer as a benchmark to systematically evaluate a model’s robustness and show that MSeg training yields substantially more robust models in comparison to training on individual datasets or naive mixing of datasets without the presented contributions. A model trained on MSeg ranks first on the WildDash leaderboard for robust semantic segmentation, with no exposure to WildDash data during training.Keywords
All Related Versions
This publication has 29 references indexed in Scilit:
- The Cityscapes Dataset for Semantic Urban Scene Understanding2016
- SUN RGB-D: A RGB-D scene understanding benchmark suite2015
- Fully convolutional networks for semantic segmentation2015
- The Role of Context for Object Detection and Semantic Segmentation in the Wild2014
- Vision meets robotics: The KITTI datasetThe International Journal of Robotics Research, 2013
- Unbiased look at dataset bias2011
- The Pascal Visual Object Classes (VOC) ChallengeInternational Journal of Computer Vision, 2009
- TextonBoost for Image Understanding: Multi-Class Object Recognition and Segmentation by Jointly Modeling Texture, Layout, and ContextInternational Journal of Computer Vision, 2007
- Improving predictive inference under covariate shift by weighting the log-likelihood functionJournal of Statistical Planning and Inference, 2000
- Sample Selection Bias as a Specification ErrorEconometrica, 1979