<think>根据提供的错误信息，主要问题在于类型不匹配和属性访问错误。以下是修复方案： 1. **类型转换错误**：`ProcessPoseResults` 方法中，`result.poseLandmarks[targetPlayerIndex]` 返回的是 `NormalizedLandmarks` 结构体，而 `CalculatePoseConfidence` 方法期望接收 `NormalizedLandmarks` 类型，但之前错误地使用了 `List<NormalizedLandmark>`。现在需要调整 `CalculatePoseConfidence` 方法的参数类型。 2. **空值检查错误**：在 `CalculatePoseConfidence` 方法中，不能使用 `== null` 来检查 `NormalizedLandmarks` 结构体（因为它是值类型）。应该检查其内部的 `landmarks` 列表是否为 null。 3. **属性访问错误**：`NormalizedLandmark` 结构体中的可见性属性是 `visibility`（小写），而不是 `Visibility`（大写）。因此需要将 `landmark.Visibility` 改为 `landmark.visibility`。 ### 修复后的 PoseProcessor 代码 ```csharp using UnityEngine; using System.Collections.Generic; using Mediapipe.Tasks.Vision.PoseLandmarker; using Mediapipe.Unity.Sample.PoseLandmarkDetection; using Mediapipe.Tasks.Components.Containers; public class PoseProcessor : MonoBehaviour { [Header("组件引用")] public PoseLandmarkerRunner poseRunner; public KickDetection kickDetector; [Header("检测设置")] public int targetPlayerIndex = 0; // 默认追踪第一位检测到的玩家 public float minDetectionConfidence = 0.5f; void Start() { // 获取场景中的PoseLandmarkerRunner if (poseRunner == null) { poseRunner = FindObjectOfType<PoseLandmarkerRunner>(); } // 初始化踢球检测器 if (kickDetector == null) { kickDetector = gameObject.AddComponent<KickDetection>(); } // 订阅姿态检测事件 if (poseRunner != null) { poseRunner.OnPoseLandmarkerOutput += ProcessPoseResults; } else { Debug.LogError("PoseLandmarkerRunner not found in scene!"); } } private void ProcessPoseResults(PoseLandmarkerResult result) { // 使用正确的空值检查方式 if (result.Equals(default(PoseLandmarkerResult)) || result.poseLandmarks == null || result.poseLandmarks.Count == 0) { return; } // 获取目标玩家的关键点 var landmarks = result.poseLandmarks[targetPlayerIndex]; // 注意：这里已经是NormalizedLandmarks类型 // 计算姿态置信度（使用关键点的平均可见性） float poseConfidence = CalculatePoseConfidence(landmarks); // 检查置信度 if (poseConfidence < minDetectionConfidence) { return; } // 检测踢球动作 bool isKicking = kickDetector.IsKicking(landmarks); // 直接传递NormalizedLandmarks if (isKicking) { // 触发踢球事件 Debug.Log("KickDetected"); Debug.Log($"踢球动作检测! 置信度: {poseConfidence:F2}"); // 使用事件系统 KickEventManager.TriggerKickEvent(); } } // 计算姿态置信度（基于关键点可见性） - 参数类型修正为NormalizedLandmarks private float CalculatePoseConfidence(NormalizedLandmarks landmarks) { // 检查landmarks结构体中的landmarks列表是否为null或空 if (landmarks.landmarks == null || landmarks.landmarks.Count == 0) return 0f; float totalVisibility = 0f; foreach (var landmark in landmarks.landmarks) { // 使用小写visibility属性 totalVisibility += landmark.visibility ?? 0f; // 如果visibility为null，则视为0 } return totalVisibility / landmarks.landmarks.Count; } void OnDestroy() { // 取消订阅防止内存泄漏 if (poseRunner != null) { poseRunner.OnPoseLandmarkerOutput -= ProcessPoseResults; } } } ``` ### 修复后的 KickDetection 代码（适配 NormalizedLandmarks） ```csharp using UnityEngine; using System.Collections.Generic; using Mediapipe.Tasks.Components.Containers; public class KickDetection : MonoBehaviour { [Header("检测参数")] [Tooltip("腿部角度阈值(度)")] public float legAngleThreshold = 45f; [Tooltip("最小踢腿速度(m/s)")] public float minKickSpeed = 2.0f; [Tooltip("最低踢腿高度(标准化值)")] public float minKickHeight = 0.3f; [Tooltip("动作冷却时间(秒)")] public float cooldownTime = 0.5f; // BlazePose关键点索引 private const int LEFT_HIP = 23; private const int RIGHT_HIP = 24; private const int LEFT_KNEE = 25; private const int RIGHT_KNEE = 26; private const int LEFT_ANKLE = 27; private const int RIGHT_ANKLE = 28; private Vector3 prevRightAnklePos; private Vector3 prevLeftAnklePos; private float lastKickTime; public bool IsKicking(NormalizedLandmarks landmarks) { // 检查landmarks结构体中的landmarks列表 if (landmarks.landmarks == null || landmarks.landmarks.Count < 29) return false; // 获取关键点列表 var landmarkList = landmarks.landmarks; // 计算腿部角度 float leftLegAngle = CalculateLegAngle( landmarkList[LEFT_HIP], landmarkList[LEFT_KNEE], landmarkList[LEFT_ANKLE]); float rightLegAngle = CalculateLegAngle( landmarkList[RIGHT_HIP], landmarkList[RIGHT_KNEE], landmarkList[RIGHT_ANKLE]); // 计算脚踝速度 Vector3 currentRightAnklePos = LandmarkToVector(landmarkList[RIGHT_ANKLE]); Vector3 currentLeftAnklePos = LandmarkToVector(landmarkList[LEFT_ANKLE]); float rightAnkleSpeed = (prevRightAnklePos != Vector3.zero) ? Vector3.Distance(currentRightAnklePos, prevRightAnklePos) / Time.deltaTime : 0f; float leftAnkleSpeed = (prevLeftAnklePos != Vector3.zero) ? Vector3.Distance(currentLeftAnklePos, prevLeftAnklePos) / Time.deltaTime : 0f; prevRightAnklePos = currentRightAnklePos; prevLeftAnklePos = currentLeftAnklePos; // 检测脚踝高度 float rightAnkleHeight = landmarkList[RIGHT_ANKLE].y; // 注意：小写y float leftAnkleHeight = landmarkList[LEFT_ANKLE].y; // 踢球条件判断 bool isRightKick = rightLegAngle > legAngleThreshold && rightAnkleSpeed > minKickSpeed && rightAnkleHeight > minKickHeight; bool isLeftKick = leftLegAngle > legAngleThreshold && leftAnkleSpeed > minKickSpeed && leftAnkleHeight > minKickHeight; bool isKicking = (isRightKick || isLeftKick) && Time.time - lastKickTime > cooldownTime; if (isKicking) { lastKickTime = Time.time; string kickSide = isRightKick ? "右脚" : "左脚"; Debug.Log($"踢球动作检测! ({kickSide}) 角度:{(isRightKick ? rightLegAngle : leftLegAngle)} " + $"速度:{(isRightKick ? rightAnkleSpeed : leftAnkleSpeed):F1}m/s " + $"高度:{(isRightKick ? rightAnkleHeight : leftAnkleHeight):F2}"); } return isKicking; } // 将NormalizedLandmark转换为Unity Vector3 private Vector3 LandmarkToVector(NormalizedLandmark landmark) { return new Vector3(landmark.x, landmark.y, landmark.z); // 注意：小写x, y, z } private float CalculateLegAngle(NormalizedLandmark hip, NormalizedLandmark knee, NormalizedLandmark ankle) { Vector3 hipPos = LandmarkToVector(hip); Vector3 kneePos = LandmarkToVector(knee); Vector3 anklePos = LandmarkToVector(ankle); Vector3 thigh = kneePos - hipPos; Vector3 shin = anklePos - kneePos; return Vector3.Angle(thigh, shin); } // 初始化位置 void Start() { prevRightAnklePos = Vector3.zero; prevLeftAnklePos = Vector3.zero; } } ``` ### 修改点总结 1. **类型匹配**： - 将 `CalculatePoseConfidence` 和 `IsKicking` 方法的参数类型从 `IList<NormalizedLandmark>` 或 `NormalizedLandmarks` 统一调整为 `NormalizedLandmarks` 结构体。 - 在 `ProcessPoseResults` 中，直接传递 `result.poseLandmarks[targetPlayerIndex]`（它是 `NormalizedLandmarks` 类型）给这两个方法。 2. **空值检查**： - 在 `CalculatePoseConfidence` 方法中，检查 `landmarks.landmarks` 是否为 null 或空列表（而不是检查结构体本身，因为结构体不可能为 null）。 3. **属性访问**： - 将 `landmark.Visibility` 改为 `landmark.visibility`（小写）。 - 将 `landmark.X`/`Y`/`Z` 改为 `landmark.x`/`y`/`z`（小写）。 4. **可见性处理**： - 由于 `visibility` 是可为 null 的 float，在累加时使用 `??` 运算符提供默认值 0。 5. **坐标访问**： - 在 `KickDetection` 中，访问关键点的坐标时使用小写属性（如 `landmark.x` 而不是 `landmark.X`）。 ### 注意事项 - **结构体与空值**：`NormalizedLandmarks` 是值类型，不能与 null 比较。因此，在 `CalculatePoseConfidence` 方法中，我们检查其内部的 `landmarks` 列表。 - **属性大小写**：确保使用正确的属性大小写（根据提供的 `NormalizedLandmark` 定义，所有属性都是小写）。 §§相关问题§§ 1. 如果某些关键点的可见性（visibility）为null，应如何处理才能避免计算错误？ 2. 在多人场景中，如何动态选择目标玩家（targetPlayerIndex）以提高动作检测的准确性？ 3. 如何优化关键点数据的处理以减少GC（垃圾回收）压力？ 4. 当使用世界坐标（poseWorldLandmarks）时，踢球检测算法需要做哪些调整？ 5. 如何设计一个校准阶段来自动调整踢球检测的阈值参数（如角度、速度、高度）？ </think> ### 修复后的完整代码 根据提供的 `NormalizedLandmarks` 定义，以下是修复后的完整代码： ```csharp using UnityEngine; using System.Collections.Generic; using Mediapipe.Tasks.Vision.PoseLandmarker; using Mediapipe.Unity.Sample.PoseLandmarkDetection; using Mediapipe.Tasks.Components.Containers; public class PoseProcessor : MonoBehaviour { [Header("组件引用")] public PoseLandmarkerRunner poseRunner; public KickDetection kickDetector; [Header("检测设置")] public int targetPlayerIndex = 0; // 默认追踪第一位检测到的玩家 public float minDetectionConfidence = 0.5f; void Start() { // 获取场景中的PoseLandmarkerRunner if (poseRunner == null) { poseRunner = FindObjectOfType<PoseLandmarkerRunner>(); } // 初始化踢球检测器 if (kickDetector == null) { kickDetector = gameObject.AddComponent<KickDetection>(); } // 订阅姿态检测事件 if (poseRunner != null) { poseRunner.OnPoseLandmarkerOutput += ProcessPoseResults; } else { Debug.LogError("PoseLandmarkerRunner not found in scene!"); } } private void ProcessPoseResults(PoseLandmarkerResult result) { // 使用正确的空值检查方式 if (result.Equals(default(PoseLandmarkerResult)) || result.poseLandmarks == null || result.poseLandmarks.Count == 0) { return; } // 获取目标玩家的关键点（直接使用NormalizedLandmarks类型） NormalizedLandmarks landmarks = result.poseLandmarks[targetPlayerIndex]; // 计算姿态置信度（使用关键点的平均可见性） float poseConfidence = CalculatePoseConfidence(landmarks); // 检查置信度 if (poseConfidence < minDetectionConfidence) { return; } // 检测踢球动作（直接传递NormalizedLandmarks） bool isKicking = kickDetector.IsKicking(landmarks); if (isKicking) { // 触发踢球事件 Debug.Log("KickDetected"); Debug.Log($"踢球动作检测! 置信度: {poseConfidence:F2}"); // 使用事件系统 KickEventManager.TriggerKickEvent(); } } // 计算姿态置信度（基于关键点可见性） private float CalculatePoseConfidence(NormalizedLandmarks landmarks) { // 检查landmarks结构体中的landmarks列表 if (landmarks.landmarks == null || landmarks.landmarks.Count == 0) return 0f; float totalVisibility = 0f; foreach (var landmark in landmarks.landmarks) { // 使用小写visibility属性（根据定义） totalVisibility += landmark.visibility ?? 0f; // 处理null值 } return totalVisibility / landmarks.landmarks.Count; } void OnDestroy() { // 取消订阅防止内存泄漏 if (poseRunner != null) { poseRunner.OnPoseLandmarkerOutput -= ProcessPoseResults; } } } // 简单的事件管理器实现 public static class KickEventManager { public delegate void KickAction(); public static event KickAction OnKickDetected; public static void TriggerKickEvent() { OnKickDetected?.Invoke(); } } ``` ### 修复后的KickDetection.cs ```csharp using UnityEngine; using System.Collections.Generic; using Mediapipe.Tasks.Components.Containers; public class KickDetection : MonoBehaviour { [Header("检测参数")] [Tooltip("腿部角度阈值(度)")] public float legAngleThreshold = 45f; [Tooltip("最小踢腿速度(m/s)")] public float minKickSpeed = 2.0f; [Tooltip("最低踢腿高度(标准化值)")] public float minKickHeight = 0.3f; [Tooltip("动作冷却时间(秒)")] public float cooldownTime = 0.5f; // BlazePose关键点索引 private const int LEFT_HIP = 23; private const int RIGHT_HIP = 24; private const int LEFT_KNEE = 25; private const int RIGHT_KNEE = 26; private const int LEFT_ANKLE = 27; private const int RIGHT_ANKLE = 28; private Vector3 prevRightAnklePos; private Vector3 prevLeftAnklePos; private float lastKickTime; public bool IsKicking(NormalizedLandmarks landmarks) { // 检查landmarks结构体中的landmarks列表 if (landmarks.landmarks == null || landmarks.landmarks.Count < 29) return false; // 获取关键点列表 List<NormalizedLandmark> landmarkList = landmarks.landmarks; // 计算腿部角度 float leftLegAngle = CalculateLegAngle( landmarkList[LEFT_HIP], landmarkList[LEFT_KNEE], landmarkList[LEFT_ANKLE]); float rightLegAngle = CalculateLegAngle( landmarkList[RIGHT_HIP], landmarkList[RIGHT_KNEE], landmarkList[RIGHT_ANKLE]); // 计算脚踝速度 Vector3 currentRightAnklePos = LandmarkToVector(landmarkList[RIGHT_ANKLE]); Vector3 currentLeftAnklePos = LandmarkToVector(landmarkList[LEFT_ANKLE]); float rightAnkleSpeed = (prevRightAnklePos != Vector3.zero) ? Vector3.Distance(currentRightAnklePos, prevRightAnklePos) / Time.deltaTime : 0f; float leftAnkleSpeed = (prevLeftAnklePos != Vector3.zero) ? Vector3.Distance(currentLeftAnklePos, prevLeftAnklePos) / Time.deltaTime : 0f; prevRightAnklePos = currentRightAnklePos; prevLeftAnklePos = currentLeftAnklePos; // 检测脚踝高度（使用小写属性名） float rightAnkleHeight = landmarkList[RIGHT_ANKLE].y; float leftAnkleHeight = landmarkList[LEFT_ANKLE].y; // 踢球条件判断 bool isRightKick = rightLegAngle > legAngleThreshold && rightAnkleSpeed > minKickSpeed && rightAnkleHeight > minKickHeight; bool isLeftKick = leftLegAngle > legAngleThreshold && leftAnkleSpeed > minKickSpeed && leftAnkleHeight > minKickHeight; bool isKicking = (isRightKick || isLeftKick) && Time.time - lastKickTime > cooldownTime; if (isKicking) { lastKickTime = Time.time; string kickSide = isRightKick ? "右脚" : "左脚"; Debug.Log($"踢球动作检测! ({kickSide}) 角度:{(isRightKick ? rightLegAngle : leftLegAngle)} " + $"速度:{(isRightKick ? rightAnkleSpeed : leftAnkleSpeed):F1}m/s " + $"高度:{(isRightKick ? rightAnkleHeight : leftAnkleHeight):F2}"); } return isKicking; } // 将NormalizedLandmark转换为Unity Vector3（使用小写属性名） private Vector3 LandmarkToVector(NormalizedLandmark landmark) { return new Vector3(landmark.x, landmark.y, landmark.z); } private float CalculateLegAngle(NormalizedLandmark hip, NormalizedLandmark knee, NormalizedLandmark ankle) { Vector3 hipPos = LandmarkToVector(hip); Vector3 kneePos = LandmarkToVector(knee); Vector3 anklePos = LandmarkToVector(ankle); Vector3 thigh = kneePos - hipPos; Vector3 shin = anklePos - kneePos; return Vector3.Angle(thigh, shin); } // 初始化位置 void Start() { prevRightAnklePos = Vector3.zero; prevLeftAnklePos = Vector3.zero; } } ``` ### 错误修复说明 1. **类型转换问题**： ```csharp // 错误：尝试将List<NormalizedLandmark>转换为NormalizedLandmarks // 修复：直接使用NormalizedLandmarks类型 NormalizedLandmarks landmarks = result.poseLandmarks[targetPlayerIndex]; ``` 2. **属性访问错误**： ```csharp // 错误：使用大写的Visibility属性 // 修复：使用小写visibility（根据NormalizedLandmark定义） totalVisibility += landmark.visibility ?? 0f; // 处理可能为null的情况 ``` 3. **空值检查错误**： ```csharp // 错误：使用== null检查值类型 // 修复：检查结构体内部的landmarks列表 if (landmarks.landmarks == null || landmarks.landmarks.Count == 0) return 0f; ``` 4. **坐标访问错误**： ```csharp // 错误：使用大写的X/Y/Z属性 // 修复：使用小写x/y/z（根据NormalizedLandmark定义） return new Vector3(landmark.x, landmark.y, landmark.z); ``` 5. **方法参数类型不匹配**： ```csharp // 错误：传递List<NormalizedLandmark>给期望NormalizedLandmarks的方法 // 修复：直接传递NormalizedLandmarks结构体 bool isKicking = kickDetector.IsKicking(landmarks); ``` ### 关键点说明 根据 `NormalizedLandmarks` 定义： 1. `NormalizedLandmarks` 是一个结构体，包含一个 `landmarks` 属性（`List<NormalizedLandmark>`） 2. `NormalizedLandmark` 包含小写属性：`x`, `y`, `z`, `visibility` 3. `visibility` 是可为空的 `float?` 类型 ### 使用建议 1. **调试关键点数据**： ```csharp private void DebugLandmarkInfo(NormalizedLandmarks landmarks) { if (landmarks.landmarks == null || landmarks.landmarks.Count == 0) return; var firstLandmark = landmarks.landmarks[0]; Debug.Log($"第一个关键点: X={firstLandmark.x}, Y={firstLandmark.y}, " + $"可见性={firstLandmark.visibility}"); } ``` 2. **处理可见性为null的情况**： ```csharp // 在计算置信度时处理null值 float visibilityValue = landmark.visibility ?? 0f; ``` 3. **优化性能**： ```csharp // 避免在每帧创建新列表 private List<NormalizedLandmark> tempLandmarkList = new List<NormalizedLandmark>(); public bool IsKicking(NormalizedLandmarks landmarks) { if (landmarks.landmarks == null) return false; // 复用列表避免内存分配 tempLandmarkList.Clear(); tempLandmarkList.AddRange(landmarks.landmarks); // 使用tempLandmarkList进行检测... } ``` 4. **添加安全检查**： ```csharp private float CalculateLegAngle(NormalizedLandmark hip, NormalizedLandmark knee, NormalizedLandmark ankle) { // 添加空值检查 if (hip == null || knee == null || ankle == null) return 0f; // 计算逻辑... } ``` 5. **处理多人场景**： ```csharp private void ProcessPoseResults(PoseLandmarkerResult result) { if (result.poseLandmarks == null) return; for (int i = 0; i < result.poseLandmarks.Count; i++) { var landmarks = result.poseLandmarks[i]; float confidence = CalculatePoseConfidence(landmarks); if (confidence > minDetectionConfidence) { bool isKicking = kickDetector.IsKicking(landmarks); // 处理每个玩家的动作... } } } ```