Update README.md
Browse files
README.md
CHANGED
|
@@ -9,14 +9,15 @@ base_model:
|
|
| 9 |
- Ultralytics/YOLO26
|
| 10 |
---
|
| 11 |
|
| 12 |
-
# Safety Detection
|
| 13 |
|
| 14 |
A fine-tuned YOLO model for detecting fire and smoke in images and video streams, built for real-time safety monitoring.
|
| 15 |
|
| 16 |
## Model Details
|
| 17 |
- **Architecture:** YOLOv26 (fine-tuned)
|
| 18 |
- **Framework:** PyTorch
|
| 19 |
-
- **
|
|
|
|
| 20 |
|
| 21 |
## Classes
|
| 22 |
| ID | Label |
|
|
@@ -27,6 +28,18 @@ A fine-tuned YOLO model for detecting fire and smoke in images and video streams
|
|
| 27 |
## Dataset
|
| 28 |
Fine-tuned on the [Home Fire Dataset](https://www.kaggle.com/datasets/pengbo00/home-fire-dataset) from Kaggle.
|
| 29 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
## Usage
|
| 31 |
```python
|
| 32 |
from ultralytics import YOLO
|
|
@@ -37,4 +50,5 @@ results = model("image.jpg")
|
|
| 37 |
|
| 38 |
## Limitations
|
| 39 |
- Trained on home fire scenarios — performance may degrade in industrial or outdoor environments
|
|
|
|
| 40 |
- Not validated for production safety-critical systems without further testing
|
|
|
|
| 9 |
- Ultralytics/YOLO26
|
| 10 |
---
|
| 11 |
|
| 12 |
+
# Safety Detection
|
| 13 |
|
| 14 |
A fine-tuned YOLO model for detecting fire and smoke in images and video streams, built for real-time safety monitoring.
|
| 15 |
|
| 16 |
## Model Details
|
| 17 |
- **Architecture:** YOLOv26 (fine-tuned)
|
| 18 |
- **Framework:** PyTorch
|
| 19 |
+
- **Epochs:** 52
|
| 20 |
+
- **Experiment Tracking:** ClearML
|
| 21 |
|
| 22 |
## Classes
|
| 23 |
| ID | Label |
|
|
|
|
| 28 |
## Dataset
|
| 29 |
Fine-tuned on the [Home Fire Dataset](https://www.kaggle.com/datasets/pengbo00/home-fire-dataset) from Kaggle.
|
| 30 |
|
| 31 |
+
|
| 32 |
+
- **Training Logs:** [ClearML Experiment](https://app.clear.ml/projects/bffe65b5fe1649dd9d202e181ba92fe0/tasks/f57871573c9d4d969dd5867004857d99/scalars)
|
| 33 |
+
|
| 34 |
+
|
| 35 |
+
## Evaluation Metrics
|
| 36 |
+
| Metric | Value |
|
| 37 |
+
|-----------|-------|
|
| 38 |
+
| mAP@50 | 0.930 |
|
| 39 |
+
| mAP@50-95 | 0.626 |
|
| 40 |
+
| Precision | 0.913 |
|
| 41 |
+
| Recall | 0.891 |
|
| 42 |
+
|
| 43 |
## Usage
|
| 44 |
```python
|
| 45 |
from ultralytics import YOLO
|
|
|
|
| 50 |
|
| 51 |
## Limitations
|
| 52 |
- Trained on home fire scenarios — performance may degrade in industrial or outdoor environments
|
| 53 |
+
- Detection confidence decreases at stricter IoU thresholds (mAP@50-95: 0.626)
|
| 54 |
- Not validated for production safety-critical systems without further testing
|