NovaSAR-Based Automated Ship Target Recognition Dataset
This repository provides benchmark experiments of deep learning models on the NASTaR dataset.
The NovaSAR Automated Ship Target Recognition (NASTaR) comprises 3415 ship patches extracted from NovaSAR S-band imagery, with labels aligned to AIS (Automatic Identification System) data. Key features of the dataset include: 23 distinct ship classes, separation between inshore and offshore samples, and an auxiliary wake dataset containing 500 patches where ship wakes are visible.
Download the dataset from the following link.
Read the full paper on IEEE Xplore from the following link.
The latest open-access version of the paper is available on arXiv here.
You can use the notebook NovaSAR_ShipIdentificator.ipynb to explore the dataset and run the provided code.
The following bar chart, pie chart, and figure illustrate the distribution of ship types for the extracted patches, including those that feature wakes, as well as some examples of these patches.
| Categories | OA | AA | APr | AF1 |
|---|---|---|---|---|
| Fishing, Other | 87.5 ± 1.6 | 74.1 ± 3.6 | 78.3 ± 3.4 | 75.6 ± 3.1 |
| Fishing, Cargo | 86.6 ± 2.2 | 84.3 ± 3.5 | 84.3 ± 2.3 | 84.2 ± 2.9 |
| Cargo, Tanker | 73.1 ± 2.4 | 65.2 ± 2.9 | 70.4 ± 3.9 | 66.0 ± 3.3 |
| Fishing, Cargo, Tanker | 67.0 ± 2.9 | 64.4 ± 3.0 | 66.2 ± 3.2 | 64.6 ± 3.0 |
| Fishing, Passenger, Cargo, Tanker | 53.0 ± 1.7 | 49.3 ± 3.0 | 53.7 ± 2.1 | 49.8 ± 2.5 |
| Categories | OA | AA | APr | AF1 |
|---|---|---|---|---|
| Fishing, Other | 87.6 ± 3.0 | 75.4 ± 4.0 | 79.5 ± 5.6 | 76.5 ± 3.8 |
| Fishing, Cargo | 86.4 ± 2.5 | 84.5 ± 3.5 | 84.0 ± 2.9 | 84.1 ± 3.1 |
| Cargo, Tanker | 73.2 ± 1.8 | 66.0 ± 1.9 | 70.2 ± 3.0 | 66.9 ± 2.0 |
| Fishing, Cargo, Tanker | 68.8 ± 2.2 | 66.4 ± 2.4 | 68.6 ± 2.7 | 66.8 ± 2.4 |
| Fishing, Passenger, Cargo, Tanker | 58.2 ± 1.8 | 54.5 ± 2.0 | 58.9 ± 2.4 | 55.5 ± 2.1 |
| Categories | OA | AA | APr | AF1 |
|---|---|---|---|---|
| Fishing, Other | 87.3 ± 1.1 | 72.5 ± 3.6 | 78.0 ± 2.3 | 74.5 ± 3.0 |
| Fishing, Cargo | 84.3 ± 2.2 | 80.3 ± 3.2 | 82.0 ± 2.6 | 81.0 ± 2.9 |
| Cargo, Tanker | 73.1 ± 2.3 | 65.7 ± 3.1 | 70.1 ± 3.4 | 66.5 ± 3.3 |
| Fishing, Cargo, Tanker | 68.3 ± 3.3 | 64.3 ± 3.4 | 68.4 ± 4.4 | 64.9 ± 3.6 |
| Fishing, Passenger, Cargo, Tanker | 57.1 ± 2.4 | 54.0 ± 2.6 | 57.0 ± 2.7 | 54.7 ± 2.5 |
| Categories | OA | AA | APr | AF1 |
|---|---|---|---|---|
| Fishing, Other | 87.6 ± 2.9 | 78.2 ± 5.5 | 79.0 ± 4.8 | 77.6 ± 4.5 |
| Fishing, Cargo | 88.5 ± 3.3 | 84.7 ± 4.9 | 87.9 ± 3.5 | 85.8 ± 4.4 |
| Cargo, Tanker | 75.4 ± 1.6 | 65.7 ± 2.1 | 76.4 ± 4.0 | 66.7 ± 2.5 |
| Fishing, Cargo, Tanker | 70.8 ± 2.4 | 67.8 ± 1.9 | 71.2 ± 3.4 | 68.5 ± 2.4 |
| Fishing, Passenger, Cargo, Tanker | 61.9 ± 1.5 | 58.1 ± 2.5 | 64.5 ± 1.8 | 59.6 ± 2.1 |
| Categories | OA | AA | APr | AF1 |
|---|---|---|---|---|
| Fishing, Other | 83.5 ± 1.6 | 76.7 ± 3.5 | 71.9 ± 2.3 | 73.4 ± 2.0 |
| Fishing, Cargo | 86.9 ± 1.1 | 83.7 ± 2.2 | 85.2 ± 1.4 | 84.3 ± 1.5 |
| Cargo, Tanker | 74.4 ± 2.0 | 65.2 ± 3.3 | 73.7 ± 3.6 | 66.0 ± 3.9 |
| Fishing, Cargo, Tanker | 66.8 ± 3.6 | 62.5 ± 3.0 | 68.0 ± 4.0 | 63.8 ± 3.7 |
| Fishing, Passenger, Cargo, Tanker | 58.0 ± 2.0 | 55.7 ± 2.3 | 57.6 ± 2.3 | 56.2 ± 2.4 |
@article{hosseiny2026nastar,
title={NASTaR: A NovaSAR-Based Automated Ship Target Recognition Dataset},
author={Hosseiny, Benyamin and Kamirul, Kamirul and Pappas, Odysseas and Achim, Alin},
journal={IEEE Geoscience and Remote Sensing Letters},
volume={23},
pages={1--5},
year={2026},
publisher={IEEE}
}