AP Calculation

[ksv87]

Why is AP considered this way? The P-R curve calculation doesn't look right

ultralytics/ultralytics/utils/metrics.py

Lines 514 to 516 in b8c90ba

	# Append sentinel values to beginning and end
	mrec = np.concatenate(([0.0], recall, [1.0]))
	mpre = np.concatenate(([1.0], precision, [0.0]))

Correctly it should be like this

    # Append sentinel values to beginning and end
    mrec = np.concatenate(([0.0], recall, [recall[-1]], [1.0]))
    mpre = np.concatenate(([1.0], precision, [0.0], [0.0]))

Calculation example https://pyimagesearch.com/2022/05/02/mean-average-precision-map-using-the-coco-evaluator/

[UltralyticsAssistant]

👋 Hello @ksv87, thank you for your interest in Ultralytics and for raising this topic 🚀! We recommend checking out the Ultralytics Docs for new users; it contains information on various usage scenarios, including both Python and CLI options.

If you suspect the AP calculation might be a 🐛 bug, could you please provide a minimum reproducible example to help us better understand the issue? This will be quite helpful for our team to reproduce and investigate the matter directly.

For a custom calculation or theoretical ❓ question, please feel free to provide insights, code snippets, or examples you have encountered, such as the ones you've shared in your message. This additional context is valuable to us. 😊

Join the Ultralytics community where you feel most comfortable. For immediate discussions, you can connect on Discord 🎧. If you prefer detailed discussions, our Discourse platform is a great choice, or explore shared experiences in our Subreddit.

Upgrade

To ensure any existing issues aren't already resolved, consider upgrading to the latest ultralytics package along with all necessary requirements in a Python>=3.8 environment, and verify compatibility with PyTorch>=1.8:

pip install -U ultralytics

Environments

You may run YOLO in these up-to-date verified environments (which come pre-installed with all dependencies including CUDA, CUDNN, Python, and PyTorch):

• Notebooks with free GPU:
• Google Cloud Deep Learning VM. Check out the GCP Quickstart Guide
• Amazon Deep Learning AMI. Refer to the AWS Quickstart Guide
• Docker Image. Consult the Docker Quickstart Guide

Status

As shown, if this badge is green, all Ultralytics CI tests are passing, confirming the correct operation of all YOLO Modes and Tasks on macOS, Windows, and Ubuntu within a 24-hour cycle and upon every commit.

This is an automated response, and an Ultralytics engineer will join the discussion shortly to assist further. 💪

[glenn-jocher]

@ksv87 thank you for your feedback. The AP calculation in our code follows the COCO evaluation metrics. For a detailed explanation, please refer to our documentation: https://docs.ultralytics.com/guides/yolo-performance-metrics/.