Mediapipe Blazepose poseclassification android 源码分析（六）

Mediapipe Blazepose poseclassification android 源码分析（六）——PoseClassifier.java

• PoseClassifier.java

    作用：Classifies Pose based on given PoseSamples. 基于pose的示例给pose分类。

    思想：KNN

    主要参数：MAX_DISTANCE_TOP_K = 30；MEAN_DISTANCE_TOP_K = 10;

    AXES_WEIGHTS = PointF3D.from(1, 1, 0.2f)因为Z轴相对于X和Y对精度影响小，所以所占权重默认0.2；

    所有参数均可人为设置

    - 方法：

        - classify(Pose pose)

public ClassificationResult classify(Pose pose) {
    return classify(extractPoseLandmarks(pose));
  }

            首先是要对传入的Pose进行extractPoseLandmarks(Pose pose)

            把Pose类转换为PointF3D的List，然后调用classify(List<PointF3D> landmarks)

        - extractPoseLandmarks(Pose pose)

  private static List<PointF3D> extractPoseLandmarks(Pose pose) {
    List<PointF3D> landmarks = new ArrayList<>();
    for (PoseLandmark poseLandmark : pose.getAllPoseLandmarks()) {
      landmarks.add(poseLandmark.getPosition3D());
    }
    return landmarks;
  }

            把Pose类转换为PointF3D的List

        - classify(List<PointF3D> landmarks)

            这部分是核心代码

public ClassificationResult classify(List<PointF3D> landmarks) {
    ClassificationRes