Buoy Detection Model (buoy-model)¶

This page documents the MHSeals buoy detection model repository, which is used to train and run object detection models for buoys using YOLOv8 and Roboflow datasets.

Overview¶

Repository: MHSeals/buoy-model
Main Language: Python
Framework: YOLOv8 (Ultralytics)
Dataset: Roboflow (customizable)

Features¶

Train custom buoy detection models using annotated datasets
Supports Roboflow integration for easy dataset management
Multiple run modes: test, detection, and tracking
Easily adaptable to other datasets by modifying code

How to Use¶

1. Clone the Repository¶

git clone https://github.com/MHSeals/buoy-model.git

2. Install Requirements¶

pip install -r requirements.txt

3. Train the Model¶

python train_annotated.py

- By default, downloads dataset from Roboflow. - To use your own dataset, modify train_annotated.py to pass your own Dataset instance and remove Roboflow-specific code.

4. Retrieve Model Weights¶

After training, weights are saved in runs/detect/<version name>/weights/best.pt.
Location may vary (project root or Python installation root).

5. Run the Model¶

Test mode:
```
python detect_test.py
```
Runs on a folder of images, allows manual navigation.
Detection mode:
```
python detect_webcam.py
```
Runs detection on webcam input.
Tracking mode:
```
python detect_tracking.py
```
Runs object tracking on video input.

Roboflow Integration¶

Dataset and model management is streamlined with Roboflow.
Download Dataset
Try Model Online

Customization & Extending¶

You can remove Roboflow-specific code to use your own datasets.
Modify train_annotated.py and related scripts for custom data pipelines.
Supports YOLOv8 features and configuration options.