Interference Polarized Ship Detection Dataset with 934 Annotated Images

934 images sized 256×256 pixels contain 2271 annotated ship targets. The dataset covers four types of typical interferences like strong sea clutter and ship ghosts, collected from six sensor types across three frequencies. It is organized according to PASCAL VOC and MS COCO standards.

Use Cases

• Train object detection models on 2271 annotated ship targets to identify vessels in maritime imagery.
• Develop algorithms to differentiate ship targets from interference types like strong sea clutter and ship ghosts.
• Benchmark model robustness using images collected from six sensor types with three different frequencies.
• Fine-tune detectors on 256×256 pixel images formatted to PASCAL VOC and MS COCO standards.

Strengths

• Includes 2271 annotated ship targets across 934 images.
• Covers four specific interference types: strong sea clutter, ship ghosts, glints, and cross-sidelobes.
• Data is collected from six sensor types with three different frequencies for diversity.
• Organized in standard formats (PASCAL VOC and MS COCO) for easier integration.

Limitations

• Dataset size of 934 images is relatively small for training large-scale deep learning models from scratch.
• Limited to a fixed image resolution of 256×256 pixels, which may not capture finer details.
• Geographic and temporal coverage of the collected real samples is not specified.

Provenance

Source

figshare, author Zhihao Cai.

Collection Method

Real samples collected from six types of sensors.

Time Range

null

Freshness

Last updated in March 2026.

Geography

null