AssertionError: negative labels:

[shliang0603]

@glenn-jocher when I train my custom dataset in yolov5 v3.0, The following error occurred：

Can you give me some advice ？

[github-actions[bot]]

Hello @shliang0603, thank you for your interest in our work! Please visit our Custom Training Tutorial to get started, and see our Jupyter Notebook , Docker Image, and Google Cloud Quickstart Guide for example environments.

If this is a bug report, please provide screenshots and minimum viable code to reproduce your issue, otherwise we can not help you.

If this is a custom model or data training question, please note Ultralytics does not provide free personal support. As a leader in vision ML and AI, we do offer professional consulting, from simple expert advice up to delivery of fully customized, end-to-end production solutions for our clients, such as:

• Cloud-based AI systems operating on hundreds of HD video streams in realtime.
• Edge AI integrated into custom iOS and Android apps for realtime 30 FPS video inference.
• Custom data training, hyperparameter evolution, and model exportation to any destination.

For more information please visit https://www.ultralytics.com.

[Aktcob]

This repo doesn't support ignore(negative label -1 means ignore) labels right now.

[shliang0603]

@Aktcob Thanks for your reply, I don't quite understand what you mean by negative label. I annotate the data set using the Labelimg tool, but the other previously annotated data sets have no problem in training

010854.txt label info:

22 0.4216666666666667 0.33 -0.7366666666666667 0.35333333333333333
22 0.415 0.6283333333333333 -0.6966666666666667 0.33666666666666667

And the annotation image 010854.jpg is as follows:

[Aktcob]

Sorry. I made a mistake.

The reason is -0.7366666666666667<0, the x1y1x2y2 should be in 0~1

[ChBrockmann]

@Aktcob You are basically correct, but I just want to tell you, that those are not x1y1x2y2 values. Those are numbers in the normalized xywh format. Meaning: x_center y_center width height For more Information you should check out: #12

[Aktcob]

@ChBrockmann Danke～

[glenn-jocher]

@Aktcob I made a new graphic to explain the normalized xywh labels. All classes must be >=0, and all box coordinates must also be 1 >= x >= 0. Does this explain it better?

I'll update the custom training tutorials with this so it's less confusing.

[shliang0603]

@Aktcob @glenn-jocher Thanks for your reply, I've got the problem.

[sp7414]

你好，我也遇到了和您一样的问题，同样报错AssertionError: negative labels: data/coco/labels/train2017/000016_4.txt，检查发现确实存在负值，请问您解决了这个问题吗？可以分享一下吗，谢谢

[github-actions[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.