Side-Scan Sonar Pipeline Detection Model Performance Metrics

Published in 2026, this dataset contains performance metrics from ablation and comparative experiments evaluating a modified YOLO11n-seg model for detecting submarine pipelines and cables in side-scan sonar imagery. The results, authored by Dandan Liu, quantify improvements in recall, mean average precision (mAP), and model parameter size. The data supports the study's findings on real-time detection and lightweight deployment for marine infrastructure.

Use Cases

• Compare bounding box recall and mAP@50-95 metrics across YOLO11n-seg, YOLO12n-seg, and YOLO13n-seg model variants.
• Analyze the impact of A2C2f and DSConv modules on instance segmentation recall and mAP@50.
• Evaluate the trade-off between detection performance (precision, integrity) and model size reduction (parameter count).

Strengths

• Quantifies specific performance gains, including a 9.7% bounding box recall improvement and a 15% reduction in model parameters.
• Results are benchmarked against established models (YOLO11n-seg, YOLO12n-seg, YOLO13n-seg) for comparative analysis.
• Data is recent, with a last updated date of April 2026.

Limitations

• The dataset is very small at 5.5 KB, indicating it contains only summary metrics, not raw imagery or training data.
• The underlying Marine-PULSE dataset's size, class distribution, and geographic origin are not described, limiting reproducibility.
• No sample data or column schema is provided, making the exact structure of the performance metrics unclear.

Provenance

Source

Dandan Liu, via figshare.

Collection Method

Results from ablation and comparative experiments on the Marine-PULSE submarine pipeline and cable dataset.

Time Range

Experiment date unspecified; results published 2026.

Freshness

Last updated April 2026.