The ABOships dataset consists of 9880 inshore and offshore images of maritime objects, belonging to one of the following categories: boat, cargoship, cruiseship, ferry, militaryship, miscboat, miscellaneous, motorboat, passengership, sailboat, seamark. Example images and definitions of these categories can be found in the reference paper. The total number of annotated objects is
41,967 and includes 9 types of ships.
The images were acquired from a collection of videos (format MPEG, with 720 p resolution at 15 FPS), extracting one image at every 15 s (or every 225 frames). The videos were acquired in the Finnish Archipelago, from a camera attached to a moving watercraft (waterbus – vesibussi in Finnish). The representation of labels for each category in the dataset is given by the following percentages, i.e. the percentage of images that include objects in each label category: seamark (37.89%), boat (20.58%), sailboat (38.88%), motorboat (41.11%), passengership (26.71%), cargoship (1.58%), ferry (9.56%), miscboat (28.30%), miscellaneous (1.30%), militaryship (25.90%), cruiseship (13.63%).
A more detailed description of the datasets and instructions can be found in the PDF description file.
- Iancu, B.; Soloviev, V.; Zelioli, L.; Lilius, J. ABOships—An Inshore and Offshore Maritime Vessel Detection Dataset with Precise Annotations. Remote Sens. 2021, 13, 988. https://doi.org/10.3390/rs13050988