# Keypoint detection evaluation

**URL:** <https://discuss.roboflow.com/t/keypoint-detection-evaluation/11447>\
**Category:** 🤝  Community Help\
**Created:** [November 12, 2025, 7:43am UTC](https://discuss.roboflow.com/t/keypoint-detection-evaluation/11447 "2025-11-12T07:43:13Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![San](https://yyz1.discourse-cdn.com/flex029/user_avatar/discuss.roboflow.com/san/32/7061_2.png) [@San](https://discuss.roboflow.com/u/San)\
**Post date:** [November 12, 2025, 7:43am UTC](https://discuss.roboflow.com/t/keypoint-detection-evaluation/11447/1 "2025-11-12T07:43:13Z")

</div>

I’m currently training a keypoint detection model using Roboflow, and I’d like to better understand how the evaluation metrics (precision, recall, OKS, mAP) are computed.

1. Does Roboflow evaluate each keypoint ( joint-by-joint)?

2. Or does it evaluate per detected person/image, aggregating all keypoints together?

And also I would like to know that, I purchased the paid plan but I can’t upload the video for prediction. Doesn’t it support for Keypoint detection model or is it limited to image uploads only?

---

<div class="post-metadata">

**Author:** ![system](https://canada1.discourse-cdn.com/flex029/uploads/roboflow1/original/1X/bb6edbfa56d0cf2ee603f061e9e09103502cc8cc.png) [@system](https://discuss.roboflow.com/u/system)\
**Post date:** [December 3, 2025, 7:43am UTC](https://discuss.roboflow.com/t/keypoint-detection-evaluation/11447/2 "2025-12-03T07:43:42Z")

</div>

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.