line_thickness and show_labels arguements don't seem to work.

[wb666greene]

I've integrated yolo8 into my security system and it is working very well. Problem is on the annotated images the thickness of the lines drawing the boxes and the object labels are too large and obscure needed details.

basically I do:
results = model.predict(image, conf=confidence-0.001, line_thickness=1,verbose=False,show_labels=False)
annotated_image = results[0].plot(line_thickness=1)

The conf= and verbose= do what i expected, while the show_labels= and line_thickness= seem to have no effect.

Any clues as to what I am misunderstanding or doing wrong here?

[wb666greene]

This part of the documentation seems incorrect in places. Playing around I got it to work with:

results = model.predict(image, conf=confidence-0.001, verbose=False)
  annotated_image = results[0].plot(line_width=1, labels=False)

[glenn-jocher]

@wb666greene hello! I'm glad to hear that YOLOv8 is working well with your security system. It looks like you've found a solution to your issue with the line_thickness and show_labels arguments. Indeed, the correct arguments to use within the plot() method are line_width for adjusting the thickness of the bounding box lines and labels to toggle the visibility of the labels.

The predict() method is used for running inference, and it doesn't directly control the visualization parameters like line_thickness or show_labels. These parameters are part of the plot() method, which is why your second snippet of code works as expected.

For future reference, you can always check the Ultralytics Docs for the most up-to-date information on the usage of different methods and their parameters. If you find any discrepancies in the documentation, please feel free to raise an issue on the GitHub repository so we can correct it. Your contributions help improve the experience for the entire YOLO community. 😊

If you have any more questions or run into further issues, don't hesitate to reach out. Happy detecting! 🚀

[dh031200]

@wb666greene

When I use save=True it works correctly, but when I plot() no arguments are passed.

model.predict(verbose=False, show_labels=False, line_width=1, save=True)

I added code to make the resulting image available even when the python program is running.

Before

results = model.predict(verbose=False)
annotated_image = results[0].plot(labels=False, line_width=1)
cv2.imwrite('output.png', annotated_image)

After

results = model.predict(verbose=False, show_labels=False, line_width=1)
annotated_image = results[0].plot()
cv2.imwrite('output.png', annotated_image)

If the code is merged, you'll be able to use it the way you thought. #2373