These tools and metrics are designed to help AI actors develop and use trustworthy AI systems and applications that respect human rights and are fair, transparent, explainable, robust, secure and safe.
3D Pose Correct Keypoints
The '3DPCK' metric (3D Pose Correct Keypoints) is a performance metric used to evaluate the accuracy of 3D human pose estimation algorithms. It measures the percentage of keypoints for which the estimated 3D pose is within a certain distance from the ground truth pose, typically expressed in millimeters.
In other words, for a given set of keypoints (e.g., joints) in a 3D human pose, the '3DPCK' metric determines the fraction of keypoints for which the estimated position is close enough to the ground truth position. The threshold distance is usually set to 150mm, which corresponds to a 5cm error in the estimated 3D position of the keypoint.
This metric is commonly used in the field of computer vision and computer graphics to evaluate the performance of algorithms that estimate the 3D pose of a human subject from 2D images or videos.
Accurate 3D pose estimation, as measured by the 3DPCK metric, supports Safety by reducing the risk of harmful or erroneous outputs in applications where human pose understanding is critical (e.g., preventing accidents in collaborative robotics or ensuring correct posture detection in healthcare). It also supports Robustness by providing a quantitative measure of system performance under various conditions, helping to ensure reliability and resilience in real-world scenarios.
References
About the metric
You can click on the links to see the associated metrics
Objective(s):
Purpose(s):
Lifecycle stage(s):
Target users:
Github stars:
- 7100
Github forks:
- 720
