Learning with Weak Annotations for Robust Maritime Obstacle Detection
Robust maritime obstacle detection is critical for safe navigation of autonomous boats and timely collision avoidance. The current state-of-the-art is based on deep segmentation networks trained on large datasets. However, per-pixel ground truth labeling of such datasets is labor-intensive and expen...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-11-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/22/23/9139 |
| _version_ | 1797462158815002624 |
|---|---|
| author | Lojze Žust Matej Kristan |
| author_facet | Lojze Žust Matej Kristan |
| author_sort | Lojze Žust |
| collection | DOAJ |
| description | Robust maritime obstacle detection is critical for safe navigation of autonomous boats and timely collision avoidance. The current state-of-the-art is based on deep segmentation networks trained on large datasets. However, per-pixel ground truth labeling of such datasets is labor-intensive and expensive. We propose a new scaffolding learning regime (SLR) that leverages weak annotations consisting of water edges, the horizon location, and obstacle bounding boxes to train segmentation-based obstacle detection networks, thereby reducing the required ground truth labeling effort by a factor of twenty. SLR trains an initial model from weak annotations and then alternates between re-estimating the segmentation pseudo-labels and improving the network parameters. Experiments show that maritime obstacle segmentation networks trained using SLR on weak annotations not only match but outperform the same networks trained with dense ground truth labels, which is a remarkable result. In addition to the increased accuracy, SLR also increases domain generalization and can be used for domain adaptation with a low manual annotation load. The SLR code and pre-trained models are freely available online. |
| first_indexed | 2024-03-09T17:32:32Z |
| format | Article |
| id | doaj.art-0ff06e563cee4c2b924fcbff0a904dab |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-09T17:32:32Z |
| publishDate | 2022-11-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-0ff06e563cee4c2b924fcbff0a904dab2023-11-24T12:09:09ZengMDPI AGSensors1424-82202022-11-012223913910.3390/s22239139Learning with Weak Annotations for Robust Maritime Obstacle DetectionLojze Žust0Matej Kristan1Faculty of Computer and Information Science, University of Ljubljana, Večna pot 113, 1000 Ljubljana, SloveniaFaculty of Computer and Information Science, University of Ljubljana, Večna pot 113, 1000 Ljubljana, SloveniaRobust maritime obstacle detection is critical for safe navigation of autonomous boats and timely collision avoidance. The current state-of-the-art is based on deep segmentation networks trained on large datasets. However, per-pixel ground truth labeling of such datasets is labor-intensive and expensive. We propose a new scaffolding learning regime (SLR) that leverages weak annotations consisting of water edges, the horizon location, and obstacle bounding boxes to train segmentation-based obstacle detection networks, thereby reducing the required ground truth labeling effort by a factor of twenty. SLR trains an initial model from weak annotations and then alternates between re-estimating the segmentation pseudo-labels and improving the network parameters. Experiments show that maritime obstacle segmentation networks trained using SLR on weak annotations not only match but outperform the same networks trained with dense ground truth labels, which is a remarkable result. In addition to the increased accuracy, SLR also increases domain generalization and can be used for domain adaptation with a low manual annotation load. The SLR code and pre-trained models are freely available online.https://www.mdpi.com/1424-8220/22/23/9139semantic segmentationweak supervisionobstacle detectionmaritime perception |
| spellingShingle | Lojze Žust Matej Kristan Learning with Weak Annotations for Robust Maritime Obstacle Detection Sensors semantic segmentation weak supervision obstacle detection maritime perception |
| title | Learning with Weak Annotations for Robust Maritime Obstacle Detection |
| title_full | Learning with Weak Annotations for Robust Maritime Obstacle Detection |
| title_fullStr | Learning with Weak Annotations for Robust Maritime Obstacle Detection |
| title_full_unstemmed | Learning with Weak Annotations for Robust Maritime Obstacle Detection |
| title_short | Learning with Weak Annotations for Robust Maritime Obstacle Detection |
| title_sort | learning with weak annotations for robust maritime obstacle detection |
| topic | semantic segmentation weak supervision obstacle detection maritime perception |
| url | https://www.mdpi.com/1424-8220/22/23/9139 |
| work_keys_str_mv | AT lojzezust learningwithweakannotationsforrobustmaritimeobstacledetection AT matejkristan learningwithweakannotationsforrobustmaritimeobstacledetection |