University of Twente Student Theses
Enhancing Vessel Re-identification with RGB-Infrared Multi-Modal Techniques
Abbema, F. van (2024) Enhancing Vessel Re-identification with RGB-Infrared Multi-Modal Techniques.
|
PDF
1MB |
| Abstract: | Vessel re-identification tries to identify a new ship as a ship that has been seen before or as an unknown ship. This field has made some good progress. However, all the literature only makes use of the visible light (RGB) modality. The use of the infrared (IR) modality has not yet been explored in this field. Using an IR modality next to the RGB modality adds new information to a sample. Exploiting this new information might result in a more generalised and expressive model and therefore a better-performing model. In this research, RGB-IR multi-modal models will be compared with the RGB-only models. In order to achieve this, a new RGB-IR vessel re-identification dataset is presented. Results show an increase of 0.023 of the weighted sum of rank-k accuracies and area under the precision-recall curve for the best RGB-IR model compared to the best RGB-only model. These results show that IR adds valuable information for vessel re-identification. |
| Item Type: | Essay (Master) |
| Clients: | Thales, Hengelo, the Netherlands |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Computer Science MSc (60300) |
| Link to this item: | https://purl.utwente.nl/essays/104655 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page