🐠 Underwater Red Circle Detection – YOLOv11x
A custom-trained YOLOv11x model for detecting red circular objects in underwater environments, such as those used in Submarine UUV (Unmanned Underwater Vehicle) navigation or robotics missions.
📦 Model Overview
This model is designed to detect red circles underwater with high precision and recall, making it ideal for use in vision systems of autonomous underwater vehicles. It is trained on a domain-specific dataset tailored for underwater visibility conditions, color distortions, and shape detection.
🧪 Training Details
- Model: YOLOv11x
- Epochs: 45
- Input Size: 640x640 pixels
- Dataset: Custom underwater dataset focused on red circle annotations
📊 Performance Metrics
| Metric | Score |
|---|---|
| Precision | 1.000 |
| Recall | 0.959 |
| [email protected] | 0.979 |
| [email protected]:0.95 | 0.930 |
| Fitness Score | 0.935 |
- Precision of 1.0 means zero false positives – all detections are valid.
- Recall of 0.959 shows the model detects nearly all actual red circles.
- High mAP scores indicate strong localization and generalization performance.
- Fitness Score balances precision and recall for robust, real-world use.
🚀 Use Cases
- Autonomous navigation for UUVs
- Underwater robotics and inspection
- Marine research and exploration tools
- Object marking and target acquisition in aquatic settings
🛠️ Example Usage
📘 Notebook coming soon: A demo notebook will be provided to show example inference on sample underwater images.
🔬 Recommendations for Future Work
- Unseen Dataset Testing: Evaluate generalization on different underwater environments.
- Real-Time Optimization: Improve inference speed for real-time deployments.
- Hyperparameter Tuning: Experiment with learning rate schedules, data augmentations, or transfer learning for potential gains.
📁 Repository
Visit the code repository:
👉 nezahatkorkmaz/submarine-circle-detection-yolov11x
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support