Vessel Detection Labeled Patches from Satellite Imagery

Validated satellite image patches for detecting military vessels, exported from a review workflow. The dataset includes image patches, metadata, and object annotations in multiple formats like CSV, JSONL, and YOLO. It was created by DefendIntelligence and last updated on 2026-05-01.

Use Cases

• Training object detection models based on annotated vessel bounding boxes.
• Evaluating model performance on satellite imagery patches containing military vessels.
• Conducting research on hard negative mining based on the inclusion of empty-label patches.
• Benchmarking maritime surveillance algorithms using the provided YOLO-format labels.

Strengths

• Includes multiple annotation formats (CSV, JSONL, YOLO) for flexibility.
• Contains 'hard negative' patches with empty labels to aid model training.
• Patches are described as 'validated/confirmed' from a specific workflow.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Row count, geographic scope, and specific image characteristics are unknown.
• License information is not provided, which may restrict usage.

Provenance

Source

DefendIntelligence

Collection Method

Exported from a military-boat-detection review workflow.

Freshness

Last updated 2026-05-01 11:50:32; freshness should be verified.