eWaSR—An Embedded-Compute-Ready Maritime Obstacle Detection Network
Maritime obstacle detection is critical for safe navigation of autonomous surface vehicles (ASVs). While the accuracy of image-based detection methods has advanced substantially, their computational and memory requirements prohibit deployment on embedded devices. In this paper, we analyze the curren...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-06-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/23/12/5386 |
| _version_ | 1797592731341553664 |
|---|---|
| author | Matija Teršek Lojze Žust Matej Kristan |
| author_facet | Matija Teršek Lojze Žust Matej Kristan |
| author_sort | Matija Teršek |
| collection | DOAJ |
| description | Maritime obstacle detection is critical for safe navigation of autonomous surface vehicles (ASVs). While the accuracy of image-based detection methods has advanced substantially, their computational and memory requirements prohibit deployment on embedded devices. In this paper, we analyze the current best-performing maritime obstacle detection network, WaSR. Based on the analysis, we then propose replacements for the most computationally intensive stages and propose its embedded-compute-ready variant, eWaSR. In particular, the new design follows the most recent advancements of transformer-based lightweight networks. eWaSR achieves comparable detection results to state-of-the-art WaSR with only a <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>0.52</mn><mo>%</mo></mrow></semantics></math></inline-formula> F1 score performance drop and outperforms other state-of-the-art embedded-ready architectures by over <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>9.74</mn><mo>%</mo></mrow></semantics></math></inline-formula> in F1 score. On a standard GPU, eWaSR runs 10× faster than the original WaSR (115 FPS vs. 11 FPS). Tests on a real embedded sensor OAK-D show that, while WaSR cannot run due to memory restrictions, eWaSR runs comfortably at 5.5 FPS. This makes eWaSR the first practical embedded-compute-ready maritime obstacle detection network. The source code and trained eWaSR models are publicly available. |
| first_indexed | 2024-03-11T01:58:17Z |
| format | Article |
| id | doaj.art-b7eb7d1976104290a5737ed6ccd9c951 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-11T01:58:17Z |
| publishDate | 2023-06-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-b7eb7d1976104290a5737ed6ccd9c9512023-11-18T12:30:17ZengMDPI AGSensors1424-82202023-06-012312538610.3390/s23125386eWaSR—An Embedded-Compute-Ready Maritime Obstacle Detection NetworkMatija Teršek0Lojze Žust1Matej Kristan2Luxonis Holding Corporation, Littleton, CO 80127, USAFaculty 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, SloveniaMaritime obstacle detection is critical for safe navigation of autonomous surface vehicles (ASVs). While the accuracy of image-based detection methods has advanced substantially, their computational and memory requirements prohibit deployment on embedded devices. In this paper, we analyze the current best-performing maritime obstacle detection network, WaSR. Based on the analysis, we then propose replacements for the most computationally intensive stages and propose its embedded-compute-ready variant, eWaSR. In particular, the new design follows the most recent advancements of transformer-based lightweight networks. eWaSR achieves comparable detection results to state-of-the-art WaSR with only a <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>0.52</mn><mo>%</mo></mrow></semantics></math></inline-formula> F1 score performance drop and outperforms other state-of-the-art embedded-ready architectures by over <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mn>9.74</mn><mo>%</mo></mrow></semantics></math></inline-formula> in F1 score. On a standard GPU, eWaSR runs 10× faster than the original WaSR (115 FPS vs. 11 FPS). Tests on a real embedded sensor OAK-D show that, while WaSR cannot run due to memory restrictions, eWaSR runs comfortably at 5.5 FPS. This makes eWaSR the first practical embedded-compute-ready maritime obstacle detection network. The source code and trained eWaSR models are publicly available.https://www.mdpi.com/1424-8220/23/12/5386maritime obstacle detectionsemantic segmentationefficient architecturelight-weight neural networkembedded hardwareOAK-D |
| spellingShingle | Matija Teršek Lojze Žust Matej Kristan eWaSR—An Embedded-Compute-Ready Maritime Obstacle Detection Network Sensors maritime obstacle detection semantic segmentation efficient architecture light-weight neural network embedded hardware OAK-D |
| title | eWaSR—An Embedded-Compute-Ready Maritime Obstacle Detection Network |
| title_full | eWaSR—An Embedded-Compute-Ready Maritime Obstacle Detection Network |
| title_fullStr | eWaSR—An Embedded-Compute-Ready Maritime Obstacle Detection Network |
| title_full_unstemmed | eWaSR—An Embedded-Compute-Ready Maritime Obstacle Detection Network |
| title_short | eWaSR—An Embedded-Compute-Ready Maritime Obstacle Detection Network |
| title_sort | ewasr an embedded compute ready maritime obstacle detection network |
| topic | maritime obstacle detection semantic segmentation efficient architecture light-weight neural network embedded hardware OAK-D |
| url | https://www.mdpi.com/1424-8220/23/12/5386 |
| work_keys_str_mv | AT matijatersek ewasranembeddedcomputereadymaritimeobstacledetectionnetwork AT lojzezust ewasranembeddedcomputereadymaritimeobstacledetectionnetwork AT matejkristan ewasranembeddedcomputereadymaritimeobstacledetectionnetwork |