| Download | - View final version: A sensor fusion platform for semantic segmentation of sea ice imagery (PDF, 4.9 MiB)
|
|---|
| Author | Search for: Duan, Richard R. J.; Search for: Scott, Andrea K.; Search for: MacMillan, Corwin G. J.; Search for: Gash, Robert M.1; Search for: Pan, Zhao |
|---|
| Affiliation | - National Research Council of Canada. Ocean, Coastal and River Engineering
|
|---|
| Format | Text, Article |
|---|
| Conference | 28th International Conference on Port and Ocean Engineering under Arctic Conditions (POAC'25), July 13-17, 2025, St. John’s, Newfoundland and Labrador, Canada |
|---|
| Subject | deep learning; image processing; sea ice navigation; semantic segmentation; sensor fusion |
|---|
| Abstract | Declining sea ice coverage in the Arctic is expected to open up routes such as the Northwest Passage for commercial shipping for greater parts of the year. However, collisions with ice are still a major risk for ships travelling through ice-covered waters. On-board ice experts can help with identifying the nearby ice conditions for navigation, but when presented with a field of different types of ice that are visually similar, it is difficult to accurately assess the ice conditions to plot a safe route. We introduce a multi-sensor system that leverages the strengths of sensor fusion to provide unique information that can highlight the differences in ice beyond what a human expert can see. Unique data of river ice was collected in February 2025 along the shores of the Ottawa and St. Lawrence rivers as an initial test of the system. A small, labelled dataset was then created to test the performance of some basic sensor fusion methods on a pre-trained image segmentation network. |
|---|
| Publication date | 2025-08 |
|---|
| Publisher | Lulea University of Technology |
|---|
| In | |
|---|
| Language | English |
|---|
| Peer reviewed | Yes |
|---|
| Export citation | Export as RIS |
|---|
| Report a correction | Report a correction (opens in a new tab) |
|---|
| Record identifier | e265a1e2-3b84-45c4-97cc-3ce0a740a4a1 |
|---|
| Record created | 2025-10-06 |
|---|
| Record modified | 2025-10-08 |
|---|
