graviti
Products
Resources
About us
Mseg
2D Box Tracking
2D Polygon
Autonomous Driving
|...
License: Unknown

Overview

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.

Citation

Please use the following citation when referencing the dataset:

@InProceedings{MSeg_2020_CVPR,
author = {Lambert, John and Zhuang, Liu and Sener, Ozan and Hays, James and Koltun, Vladlen},
title = {{MSeg}: A Composite Dataset for Multi-domain Semantic Segmentation},
booktitle = {Computer Vision and Pattern Recognition (CVPR)},
year = {2020}
}
Data Summary
Type
Image,
Amount
80K
Size
--
Provided by
John Lambert
A Ph.D. student at Georgia Tech, completed Bachelor’s and Master’s degrees in Computer Science at Stanford University in 2018, specializing in artificial intelligence.
| Amount 80K | Size --
Mseg
2D Box Tracking 2D Polygon
Autonomous Driving
License: Unknown

Overview

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.

Citation

Please use the following citation when referencing the dataset:

@InProceedings{MSeg_2020_CVPR,
author = {Lambert, John and Zhuang, Liu and Sener, Ozan and Hays, James and Koltun, Vladlen},
title = {{MSeg}: A Composite Dataset for Multi-domain Semantic Segmentation},
booktitle = {Computer Vision and Pattern Recognition (CVPR)},
year = {2020}
}
0
Start building your AI now
graviti
wechat-QR
Long pressing the QR code to follow wechat official account

Copyright@Graviti