Samyukta31/sonar_yolo

---

library_name: ultralytics

tags:

- object-detection

- yolo

- sonar

- side-scan-sonar

- marine-debris

- underwater

- computer-vision

- pytorch

- onnx

---

# SONARINTEL — Side-Scan Sonar Object Detection

YOLO11M object detection model trained for detecting underwater objects and anomalies in **Side-Scan Sonar (SSS)** imagery.

The model is part of the **SONARINTEL** project, an AI-powered system for automated underwater marine debris and anomaly detection using side-scan sonar imagery.

## Model Details

- **Architecture:** YOLO11M

- **Task:** Object Detection

- **Input Size:** 640 × 640

- **Number of Classes:** 5

- **Training Experiment:** EXP-01

- **Framework:** Ultralytics YOLO

- **Ultralytics Version:** 8.4.142

## Classes

| ID | Class |

|---:|---|

| 0 | crab\_pot |

| 1 | submarine\_pipeline |

| 2 | shipwreck |

| 3 | ghost\_net |

| 4 | mine\_like\_contact |

## Validation Performance

Best validation performance based on **mAP@50–95**:

| Metric | Score |

|---|---:|

| Precision | 0.8275 |

| Recall | 0.7286 |

| mAP@50 | 0.6736 |

| mAP@50–95 | 0.5816 |

## Available Model Variants

### PyTorch

best.pt

Original Ultralytics PyTorch model suitable for inference and further development using the Ultralytics framework.

### ONNX FP32

best\_fp32.onnx

Full-precision ONNX model for deployment with ONNX-compatible inference runtimes.

### ONNX FP16

best\_fp16.onnx

Half-precision ONNX model intended for deployment environments supporting FP16 inference.

## Intended Use

This model is intended for research and development involving:

- Side-scan sonar image analysis

- Underwater object detection

- Marine debris detection

- Detection of ghost fishing nets

- Shipwreck detection

- Submarine pipeline detection

- Sonar anomaly detection

- Automated underwater inspection

## Limitations

Performance may vary depending on sonar sensor characteristics, imaging conditions, seabed characteristics, acquisition geometry, resolution, and the distribution of objects encountered in real-world deployments.

The mine\_like\_contact class represents sonar contacts with mine-like characteristics and should **not** be interpreted as confirmation of an actual explosive device.

This model should not be used as the sole basis for safety-critical, navigation-critical, or explosive-ordnance decisions.

## Inference

### Ultralytics

from ultralytics import YOLO



model = YOLO("best.pt")



results = model("sonar\_image.jpg")



for result in results:

    result.show()