Fix confidence threshold for ClearML debug images

[HighMans]

The confidence is converted to a percentage on line 144, but it is being compared to a default conf_threshold value of a decimal value instead of percent value.

Signed-off-by: HighMans [email protected]

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Improved variable naming and fixed confidence threshold logic in ClearML image logging.

📊 Key Changes

• Renamed the confidence variable to confidence_percentage for better clarity.
• Corrected the threshold comparison to use the raw confidence value (conf) instead of the percentage.

🎯 Purpose & Impact

• 👍 Enhances code readability by making variable names more descriptive.
• 🐞 Fixes an issue where the confidence percentage was incorrectly used in the threshold comparison, ensuring only relevant detections are logged. This can lead to more accurate debugging and visualization in ClearML.

[github-actions]

👋 Hello @HighMans, thank you for submitting a YOLOv5 🚀 PR! To allow your work to be integrated as seamlessly as possible, we advise you to:

• ✅ Verify your PR is up-to-date with upstream/master. If your PR is behind upstream/master an automatic GitHub Actions merge may be attempted by writing /rebase in a new comment, or by running the following code, replacing 'feature' with the name of your local branch:

git remote add upstream https://github.com/ultralytics/yolov5.git
git fetch upstream
# git checkout feature  # <--- replace 'feature' with local branch name
git merge upstream/master
git push -u origin -f

• ✅ Verify all Continuous Integration (CI) checks are passing.
• ✅ Reduce changes to the absolute minimum required for your bug fix or feature addition. "It is not daily increase but daily decrease, hack away the unessential. The closer to the source, the less wastage there is." -Bruce Lee

[AyushExel]

Thanks for the PR!
I think this makes sense as the confidence is being compared after being multiplied by 100. Right @thepycoder
?

yolov5/utils/loggers/clearml/clearml_utils.py

Line 144 in cff9717

confidence = round(float(conf) * 100, 2)

[HighMans]

Thanks for the PR! I think this makes sense as the confidence is being compared after being multiplied by 100. Right @thepycoder ?

yolov5/utils/loggers/clearml/clearml_utils.py

Line 144 in cff9717

confidence = round(float(conf) * 100, 2)

The confidence is converted to a percentage here:

yolov5/utils/loggers/clearml/clearml_utils.py

Line 144 in cff9717

confidence = round(float(conf) * 100, 2)

But then it's being compared two lines after here to conf_threshold which by default is .25 and not 25%:

yolov5/utils/loggers/clearml/clearml_utils.py

Line 147 in cff9717

if confidence > conf_threshold:

yolov5/utils/loggers/clearml/clearml_utils.py

Line 125 in cff9717

def log_image_with_boxes(self, image_path, boxes, class_names, image, conf_threshold=0.25):

[AyushExel]

Yup. Makes sense

[thepycoder]

Hey nice catch! That was sloppy coding from my part.

That said, don't you think it makes more sense to convert the percentage to decimal and keep the threshold like it is? I think this is pretty much default when passing along a threshold right? That it is between 0 and 1?

If you agree, can you leave the argument as it is, but just set the confidence to a decimal instead of a percentage? Then inside the f-string we can multiply it by 100 for visualisation, but it would have no effect on the original parameter which can then be properly compared to the threshold :)

[AyushExel]

@thepycoder yes I think it makes most sense to just keep the decimal and not convert it to percentage.

[HighMans]

That makes a lot more sense, should be fixed now.

yolov5/utils/loggers/clearml/clearml_utils.py

Lines 143 to 149 in 6d40357

	class_name = class_names[int(class_nr)]
	confidence_percentage = round(float(conf) * 100, 2)
	label = f"{class_name}: {confidence_percentage}%"
	if conf > conf_threshold:
	annotator.rectangle(box.cpu().numpy(), outline=color)
	annotator.box_label(box.cpu().numpy(), label=label, color=color)

[thepycoder]

That makes a lot more sense, should be fixed now.

yolov5/utils/loggers/clearml/clearml_utils.py

Lines 143 to 149 in 6d40357

	class_name = class_names[int(class_nr)]
	confidence_percentage = round(float(conf) * 100, 2)
	label = f"{class_name}: {confidence_percentage}%"
	if conf > conf_threshold:
	annotator.rectangle(box.cpu().numpy(), outline=color)
	annotator.box_label(box.cpu().numpy(), label=label, color=color)

Lgtm! Thank a lot :)

[glenn-jocher]

@HighMans @thepycoder @AyushExel PR is merged. Thank you for your contributions to YOLOv5 🚀 and Vision AI ⭐