RoadDefects-ISeg: A Dataset for Instance Segmentation of Road Surface Defects

Overview RoadDefects-ISeg is a dataset for instance segmentation of road surface defects. It contains 1000 images across 4 classes (250 images each):- Crack- Lane- Pothole- Speed_Breaker The dataset is split into train (80%), validation (10%), and test (10%) subsets. Dataset Structure RoadDefects-ISeg/│├── data.yaml # YOLO dataset configuration├── README.md # Dataset description├── LICENSE # CC BY 4.0 license│├── train/│ ├── images/ # Training images│ └── labels/ # YOLO-format segmentation labels│├── val/│ ├── images/ # Validation images│ └── labels/│└── test/ ├── images/ # Test images └── labels/ Notes The original datasets included a mix of object detection and segmentation annotations, which could not be directly used. All images were manually re-annotated to provide precise instance segmentation labels suitable for this project. This dataset is intended for research and academic use only. Acknowledgments We thank the authors of the original datasets for making their data available under the CC BY 4.0 license, which enabled the creation of this dataset: My Road Crack Dataset – Roboflow Universe Speedbreaker Final Dataset – Roboflow Universe Pothole Segmentation for Road Damage Assessment – Kaggle Lane Dataset – Roboflow Universe