🛣️ QuickFix It - Pothole Detection & Repair Verification Model (YOLOv8)
Fine-tuned YOLOv8n model trained specifically for detecting potholes and verifying road repairs on Indian roads as part of the QuickFix It civic infrastructure project.
📊 Model Information
- Architecture: YOLOv8n (Ultralytics)
- Task: Object Detection (
pothole, road damage verification) - Model Size: ~6.2 MB
- Framework: PyTorch / Ultralytics
🗃️ Training Dataset
Trained on a combined dataset of 2,281 images:
- Roboflow Indian Pothole Dataset: ~1,243 real-world Indian road scenes with variable lighting and asphalt conditions.
- Custom Before / After Repair Dataset: 501 pairs (1,002 images) covering damaged potholes and repaired/patched asphalt surfaces.
- Training Duration: 25 epochs with GPU acceleration & mixed precision.
🚀 How to Use
1. With Ultralytics (Python)
from ultralytics import YOLO
from huggingface_hub import hf_hub_download
# Download weights from Hugging Face Hub
model_path = hf_hub_download(repo_id="RoxieRoller/QuickFixIt-model", filename="best.pt")
# Load model
model = YOLO(model_path)
# Run inference
results = model.predict(source="road_image.jpg", conf=0.25)
results[0].show()
2. Integration in QuickFix It Backend
Set in your .env:
HF_MODEL_REPO=RoxieRoller/QuickFixIt-model
👥 Authors & Project
Developed for QuickFix It by Mohammed Hussain (GitHub: @mohammedhussain06).