peterhdd/pothole-detection-yolov8

YOLOv8 Pothole Detection Model

This model detects potholes in road images using YOLOv8s trained on the following dataset:

Pothole dataset

Model Details

• Architecture: YOLOv8s (Ultralytics)
• Task: Object Detection
• Classes: 1 (pothole)
• Epochs: 100
• Hardware: NVIDIA L40S (Nebius Cloud)
• Model Size: 22.5 MB

---

Sample Prediction

Training Results

Confusion Matrix

---

Usage Example (Python)

from ultralytics import YOLO

model = YOLO("https://huggingface.co/peterhdd/pothole-detection-yolov8/resolve/main/best.pt")
results = model("your_image.jpg")
results.show()

This model is part of a complete end-to-end pothole detection system including training, GPU inference, and a mobile application.

The repository can be found here:

https://github.com/PeterHdd/pothole-detection-yolo