Update README.md
Browse files
README.md
CHANGED
|
@@ -9,4 +9,20 @@ app_file: app.py
|
|
| 9 |
pinned: false
|
| 10 |
---
|
| 11 |
|
| 12 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
pinned: false
|
| 10 |
---
|
| 11 |
|
| 12 |
+
# Traffic Sign Detection
|
| 13 |
+
|
| 14 |
+
This project features a YOLOv8-based model for detecting and classifying traffic signs in images, integrated with a Gradio interface for interactive use. Upload an image to view detected traffic signs with bounding boxes and labels, powered by a model trained on the German Traffic Sign Recognition Benchmark (GTSRB) dataset.
|
| 15 |
+
|
| 16 |
+
## Usage
|
| 17 |
+
|
| 18 |
+
1. **Upload an Image**: Use the Gradio interface to upload an image containing traffic signs.
|
| 19 |
+
2. **View Results**: The app displays the image with bounding boxes and labels for detected traffic signs.
|
| 20 |
+
3. **Confidence Threshold**: Detections are filtered at a confidence score of 0.5 (adjustable in `app.py` if needed).
|
| 21 |
+
|
| 22 |
+
## Model Details
|
| 23 |
+
|
| 24 |
+
- **Architecture**: YOLOv8 nano
|
| 25 |
+
- **Dataset**: GTSRB, 43 classes, ~39,209 training images
|
| 26 |
+
- **Training**: 50 epochs, 640x640 image size, 80% training / 20% validation split
|
| 27 |
+
|
| 28 |
+
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|