datasets/segment/crack-seg/

[giscus[bot]]

datasets/segment/crack-seg/

Explore the Crack Segmentation using Ultralytics YOLOv8 Dataset, a large-scale benchmark for road safety analysis, and learn how to train a YOLO model using it.

https://docs.ultralytics.com/datasets/segment/crack-seg/

[mzg0108]

What was the performance of Yolo 8 on this dataset?
What was the mean IOU on test set?

[glenn-jocher]

Hey there!

Thanks for reaching out with your question about the YOLOv8 performance on the Crack Segmentation dataset. 🚀 For this specific dataset, we focused on optimizing the segmentation capabilities for detecting cracks in various environments. While we do not have a published mean IOU (Intersection Over Union) for the test set at the moment, our internal benchmarks showed promising results, aligning with YOLOv8's reputation for high performance across different datasets.

For an example on how to train a YOLOv8 model on the crack segmentation dataset, you might find this snippet useful:

from ultralytics import YOLO

# Load a pretrained model (recommended for training)
model = YOLO('yolov8n-seg.pt')  

# Train the model on the crack-seg dataset
results = model.train(data='crack-seg.yaml', epochs=100, imgsz=640)

If you're specifically interested in evaluating the model's performance in terms of mean IOU or any other metrics, you can use the val mode to validate your model on the testing set after training.

Feel free to dive into your own experiments and share your findings. The community and we are always eager to learn about different performance insights across various datasets! 🌟

Happy detecting, and let us know if you have any more questions or if there's anything else we can help you with!