|
--- |
|
license: mit |
|
pipeline_tag: object-detection |
|
--- |
|
# yolov8m_flying_objects_detection |
|
`yolov8m_flying_objects_detection` is a deep learning model designed to detect various flying objects, including drones, airplanes, helicopters, and birds. Based on the YOLOv8 architecture, this model provides a strong balance of speed and accuracy, making it suitable for real-time aerial surveillance and monitoring applications. |
|
|
|
## Model Summary |
|
This model has been trained to identify the following classes: |
|
- `Drone` (UAV copter) |
|
- `Airplane` |
|
- `Helicopter` |
|
- `Bird` |
|
- `Background` (no object) |
|
|
|
## Classes and Objects |
|
The model has been trained to detect and classify the following types of flying objects: |
|
|
|
1. **Drones** |
|
- DJI Matrice 200 |
|
- DJI Phantom 2 |
|
- DJI Phantom 3 |
|
- Shahed |
|
2. **Airplanes** |
|
- Airbus A220 |
|
- Airbus A220 (with stowed landing gear) |
|
- Airbus A380 |
|
- Boeing 787 |
|
- Boeing 787 (with stowed landing gear) |
|
3. **Helicopters** |
|
- Bell 407 |
|
- Robinson R44 |
|
4. **Birds** |
|
- Chayka (Seagull) |
|
- Golub (Pigeon) |
|
5. **Background** |
|
- Areas with no relevant objects. |
|
|
|
This breakdown provides more specific information on each class, helping users understand the diversity of objects the model can detect. |
|
|
|
## Confusion Matrix Analysis |
|
The confusion matrix above shows the normalized detection accuracy across different classes. Key insights include: |
|
|
|
- **Drone Detection**: 85% accurate, with occasional misclassifications as background. |
|
- **Airplane Detection**: Excellent accuracy of 99%. |
|
- **Helicopter Detection**: Correctly identified 67% of the time, with some confusion with birds. |
|
- **Bird Detection**: 68% accurate, with some misclassifications as helicopters. |
|
- **Background**: Some non-object areas are occasionally detected as objects. |
|
|
|
## Applications |
|
This model is particularly useful in scenarios where real-time identification of airborne objects is essential. Potential applications include: |
|
- **Airport Surveillance**: Detecting drones and birds to prevent collisions and ensure safety. |
|
- **Military and Security Operations**: Monitoring restricted airspaces for unauthorized drones or other aerial vehicles. |
|
- **Wildlife Monitoring**: Identifying bird movements to support ecological studies and prevent hazards. |
|
|
|
## Model Usage |
|
To use the model, follow these steps: |
|
|
|
### Install Dependencies |
|
Install the required packages listed in `requirements.txt`: |
|
```bash |
|
pip install -r requirements.txt |
|
``` |
|
|
|
## Run Inference |
|
Load the model and run inference on images or video frames using the sample `inference.py` script: |
|
|
|
```python |
|
from yolov8 import YOLO |
|
model = YOLO("yolov8m_fly_obj_detection.pt") |
|
results = model.predict("image.jpg") |
|
``` |
|
|
|
The model outputs bounding boxes for each detected object, along with their respective class labels and confidence scores. |
|
|
|
## Example Results |
|
|
|
| Class | True Positive Rate | Common Misclassifications | |
|
|------------|--------------------|---------------------------| |
|
| Drone | 85% | Background | |
|
| Airplane | 99% | None | |
|
| Helicopter | 67% | Bird | |
|
| Bird | 68% | Helicopter | |
|
|
|
## Limitations |
|
- **Class Confusion**: Some confusion exists between similar classes (e.g., helicopters and birds). |
|
- **Background Misclassification**: Non-object areas may occasionally be misclassified as objects. |
|
|
|
## License |
|
This model is released under the MIT License. Feel free to use, modify, and distribute it, but please provide proper attribution. |
|
|
|
## Citation |
|
If you use this model in your work, please consider citing it as follows: |
|
|
|
```bibtex |
|
@model{yolov8m_flying_objects_detection, |
|
title={YOLOv8m Flying Object Detection}, |
|
author={Javvanny}, |
|
year={2024}, |
|
howpublished={\url{https://huggingface.co/Javvanny/yolov8m_flying_objects_detection}}, |
|
} |
|
``` |
