Marine Obstacle Detection Dataset
Overview
This dataset contains marine videos, captured by unmanned surface vehicle (USV). The challenge, posed by this dataset, is to segment each image into three natural regions: the sky, the shore and the sea, and furthermore, detect obstacles in the sea area.
This page includes:
- Marine Obstacle Detection Dataset (MODD) v1.0
- Twelve fully-annotated videos for quantitative analysis.
- The set of images and annotations used for learning the hyper priors.
- Additional four non-annotated videos are provided for qualitative analysis.
- Matlab routines and demo for automatic performance evaluation. See readme for usage.
- Videos for ground truth visualization.
- A reference Matlab implementation of the Semantic segmentation model.
- Marine Obstacle Detection Dataset (MODD) v2.0
- A new dataset with stereo images and more challenging scenarios. See the links at the bottom of the page
Supplementary material
Here is a link to the page containing supplementary material for the paper in which the dataset was published. The page also contains a video of the USV.
The semantic segmentation model
 |
This is a Matlab demo code for the semantic segmentation model for obstacle image map estimation for unmanned surface vehicles. The demo requires downloading the MOD dataset and has pretrained hiperparameters on the MOD dataset.
|
Citing the MODD v1.0 dataset
If you use the MODD dataset, please cite the following paper in which the dataset was originally published in the form described here.
@article{Kristan2015,
title = {Fast image-based obstacle detection from unmanned surface vehicles},
author = {Matej Kristan, Vildana Sulic, Stanislav Kovacic, and Janez Perš},
year = {2015},
journal= {IEEE Transactions on Cybernetics}
}
Modd v2.0 dataset [NEW!]
Multimodal marine obstacle detection dataset (MODD v2) - This dataset contains marine videos, captured by unmanned surface vehicle (USV). The challenge, posed by this dataset, is to segment each image into three natural regions: the sky, the shore and the sea, and furthermore, detect obstacles in the sea area.
MaSTr1325 dataset [NEW!]
The MaSTr1325 dataset for training deep USV obstacle detection models (MaSTr1325) is a new large-scale marine semantic segmentation training dataset tailored for development of obstacle detection methods in small-sized coastal USVs. The dataset contains 1325 diverse images captured over a two-year span with a real USV, covering a range of realistic conditions encountered in a coastal surveillance task. All images are per-pixel semantically labelled and synchronized with inertial measurements of the on-board sensors. In addition, a dataset augmentation protocol is proposed to address slight appearance differences of the images in the training set and those in deployment.
Citing Modd v2.0
@article{Bovcon2018a,
title = {Stereo obstacle detection for unmanned surface vehicles by IMU-assisted semantic segmentation},
author = {Borja Bovcon and Rok Mandeljc and Janez Per\v{s} and Matej Kristan},
year = {2018},
journal= {Robotics and Autonomous Systems}
}