pothole-detection

Pothole Detection using Deep Learning

This repository contains trained object detection models for pothole detection as part of a bachelor’s thesis project.

The repository is hosted on Hugging Face, with the corresponding source code and experiments available on GitHub:

👉 GitHub repository: https://github.com/DanielsStulpe/pothole-detection

---

📌 Overview

The project evaluates multiple deep learning architectures for automatic pothole detection in road surface images.

Models evaluated:

• YOLOv8
• YOLO11
• YOLO26
• Faster R-CNN
• RetinaNet

A final optimized YOLO26 model is also included.

---

🏆 Best Model

• Model: YOLO26 (optimized)
• File: yolo26_best.pt

This model achieved the best overall performance across all experiments and hyperparameter optimizations.

---

📁 Model List

🔹 Baseline models

• baseline_yolov8.pt
• baseline_yolo11.pt
• baseline_yolo26.pt
• baseline_fasterrcnn.pt
• baseline_retinanet.pt

🏆 Final model

• yolo26_best.pt

---

📊 Task

• Object Detection
• Single-class (pothole detection)

---

📦 Usage (Ultralytics YOLO)

from ultralytics import YOLO

model = YOLO("yolo26_best.pt")
results = model("image.jpg")
results.show()

---

📚 Dataset

Dataset used in this project is publicly available on Roboflow:

https://universe.roboflow.com/bak-6b12k/pothole-idtb2

---

📄 License

This project is licensed under the MIT License.

MIT License

Copyright (c) 2026 Daniels Stulpe

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND.

---

🧠 Notes

This model is intended for research and academic purposes. Performance may vary depending on environmental conditions, dataset distribution, and image quality.