• 日期: 2026-02-11 20:14:22
  • 标签: 常规

基于KNN分类器的人体姿势识别算法

本算法利用KNN分类器对输入视频中的人体姿势进行检测和分类，输出每一帧的动作类型，并绘制头部关键点。

算法流程

  1. 初始化MediaPipe的人体姿势模型。
  2. 加载动作数据集，并将数据集合并为一个大的数据集。
  3. 训练KNN分类器。
  4. 处理视频文件中的每一帧： a. 读取一帧。 b. 将帧转换为RGB格式。 c. 处理人体姿势检测。 d. 判断是否检测到人体。 e. 获取关键点的信息，并计算角度。 f. 将数据输入KNN分类器进行预测。 g. 在输出图片上显示角度值和动作类型。 h. 绘制头部关键点。 i. 将帧写入输出视频文件。 j. 显示当前帧的结果。 k. 检测是否按下q键退出。
  5. 释放资源。

复杂度分析

时间复杂度： 算法的时间复杂度主要来自于人体姿势检测和KNN分类器的训练和预测过程，其中人体姿势检测的时间复杂度为O(n)，KNN分类器的训练和预测的时间复杂度为O(m^2)，其中n为关键点的数量，m为数据集的大小。因此，算法的时间复杂度为O(nm^2)。

空间复杂度： 算法的空间复杂度主要来自于存储动作数据集和KNN分类器，其中动作数据集的空间复杂度为O(mn)，KNN分类器的空间复杂度为O(m^2)，其中n为关键点的数量，m为数据集的大小。因此，算法的空间复杂度为O(mn + m^2)。

代码示例

import cv2
import mediapipe as mp
import math
import pandas as pd
from sklearn.neighbors import KNeighborsClassifier
i=0
# 初始化MediaPipe的人体姿势模型
mp_drawing = mp.solutions.drawing_utils
mp_pose = mp.solutions.pose

# 打开输入视频文件
cap = cv2.VideoCapture('9.mp4')

# 获取输入视频的帧率和分辨率
fps = int(cap.get(cv2.CAP_PROP_FPS))
width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))

# 创建输出视频文件
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
out = cv2.VideoWriter('9_1.mp4', fourcc, fps, (width, height))

# 加载动作数据集
warmup_df = pd.read_csv('dataset_warm-up.csv')
combat_df = pd.read_csv('dataset_SHIZHAN POSE.csv')
attack_df = pd.read_csv('dataset_hit.csv')
respect_df = pd.read_csv('dataset_respect.csv')
gongbu_df = pd.read_csv('dataset_gongbu.csv')

# 将数据集合并为一个大的数据集
data = pd.concat([warmup_df, combat_df, attack_df, respect_df,gongbu_df], ignore_index=True)

# 训练KNN分类器
knn = KNeighborsClassifier(n_neighbors=3)
knn.fit(data[['angle1','angle2','angle3', 'angle4','angle5' ,'angle6', "angle7","angle8","angle9","angle10","angle11"]], data['label'])

# 处理视频文件中的每一帧
with mp_pose.Pose(min_detection_confidence=0.5, min_tracking_confidence=0.5) as pose:
    while cap.isOpened():
        # 读取一帧
        ret, frame = cap.read()
        if not ret:
            break

        # 将帧转换为RGB格式
        image = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)

        # 处理人体姿势检测
        results = pose.process(image)

        # 判断是否检测到人体
        if results.pose_landmarks:
            # 绘制人体骨架
            mp_drawing.draw_landmarks(
                frame, results.pose_landmarks, mp_pose.POSE_CONNECTIONS)

            # 获取右膝、右踝和右手腕关键点的信息
            right_knee = results.pose_landmarks.landmark[mp_pose.PoseLandmark.RIGHT_KNEE]
            right_ankle = results.pose_landmarks.landmark[mp_pose.PoseLandmark.RIGHT_ANKLE]
            right_wrist = results.pose_landmarks.landmark[mp_pose.PoseLandmark.RIGHT_WRIST]

            # 获取左肩、左肘和左手腕关键点的信息
            left_shoulder = results.pose_landmarks.landmark[mp_pose.PoseLandmark.LEFT_SHOULDER]
            left_elbow = results.pose_landmarks.landmark[mp_pose.PoseLandmark.LEFT_ELBOW]
            left_wrist = results.pose_landmarks.landmark[mp_pose.PoseLandmark.LEFT_WRIST]

            # 获取右肩、右肘和右手腕关键点的信息
            right_shoulder = results.pose_landmarks.landmark[mp_pose.PoseLandmark.RIGHT_SHOULDER]
            right_elbow = results.pose_landmarks.landmark[mp_pose.PoseLandmark.RIGHT_ELBOW]
            right_wrist = results.pose_landmarks.landmark[mp_pose.PoseLandmark.RIGHT_WRIST]

            # 获取左臀、左膝和左踝关键点的信息
            left_hip = results.pose_landmarks.landmark[mp_pose.PoseLandmark.LEFT_HIP]
            left_knee = results.pose_landmarks.landmark[mp_pose.PoseLandmark.LEFT_KNEE]
            left_ankle = results.pose_landmarks.landmark[mp_pose.PoseLandmark.LEFT_ANKLE]

            # 获取右臀、右膝和右踝关键点的信息
            right_hip = results.pose_landmarks.landmark[mp_pose.PoseLandmark.RIGHT_HIP]
            right_knee = results.pose_landmarks.landmark[mp_pose.PoseLandmark.RIGHT_KNEE]
            right_ankle = results.pose_landmarks.landmark[mp_pose.PoseLandmark.RIGHT_ANKLE]

            # 获取左肩、左肘和左手腕关键点的信息 new
            left_shoulder = results.pose_landmarks.landmark[mp_pose.PoseLandmark.LEFT_SHOULDER]
            left_elbow = results.pose_landmarks.landmark[mp_pose.PoseLandmark.LEFT_ELBOW]
            left_wrist = results.pose_landmarks.landmark[mp_pose.PoseLandmark.LEFT_WRIST]

            # 计算腿与右手的角度
            angle = math.degrees(math.atan2(right_wrist.y - right_ankle.y, right_wrist.x - right_ankle.x) -
                                 math.atan2(right_knee.y - right_ankle.y, right_knee.x - right_ankle.x))

            # 获取左肩、左肘和左手腕
            angle1 = math.degrees(math.atan2(right_wrist.y - right_ankle.y, right_wrist.x - right_ankle.x) -
                                  math.atan2(right_knee.y - right_ankle.y, right_knee.x - right_ankle.x))

            # 获取左臀、左膝和左踝
            angle_dl = math.degrees(math.atan2(left_ankle.y - left_knee.y, left_ankle.x - left_knee.x) -
                                    math.atan2(left_hip.y - left_knee.y, left_hip.x - left_knee.x))

            # 获取右臀、右膝和右踝
            angle_dr = math.degrees(math.atan2(right_ankle.y - right_knee.y, right_ankle.x - right_knee.x) -
                                    math.atan2(right_hip.y - right_knee.y, right_hip.x - right_knee.x))

            # 获取右肩、右肘和右手腕
            angle_tr = math.degrees(math.atan2(right_wrist.y - right_elbow.y, right_wrist.x - right_elbow.x) -
                                    math.atan2(right_shoulder.y - right_elbow.y, right_shoulder.x - right_elbow.x))

            # 获取左肩、左肘和左手腕 new
            angle_tl = math.degrees(math.atan2(left_wrist.y - left_elbow.y, left_wrist.x - left_elbow.x) -
                                    math.atan2(left_shoulder.y - left_elbow.y, left_shoulder.x - left_elbow.x))


            ##############
            #获取左髋、左膝和左踝关键点的信息
            left_hip = results.pose_landmarks.landmark[mp_pose.PoseLandmark.LEFT_HIP]
            left_knee = results.pose_landmarks.landmark[mp_pose.PoseLandmark.LEFT_KNEE]
            left_ankle = results.pose_landmarks.landmark[mp_pose.PoseLandmark.LEFT_ANKLE]

            # 获取左髋、左膝和左踝的角度
            angle_lka = math.degrees(math.atan2(left_ankle.y - left_knee.y, left_ankle.x - left_knee.x) -
                                     math.atan2(left_hip.y - left_knee.y, left_hip.x - left_knee.x))

            # 获取腰部、左髋和左肩的角度
            waist = results.pose_landmarks.landmark[12]
            angle_hls = math.degrees(math.atan2(left_shoulder.y - waist.y, left_shoulder.x - waist.x) -
                                     math.atan2(left_hip.y - waist.y, left_hip.x - waist.x))

            # 获取右髋、右膝和右踝关键点的信息
            right_hip = results.pose_landmarks.landmark[mp_pose.PoseLandmark.RIGHT_HIP]
            right_knee = results.pose_landmarks.landmark[mp_pose.PoseLandmark.RIGHT_KNEE]
            right_ankle = results.pose_landmarks.landmark[mp_pose.PoseLandmark.RIGHT_ANKLE]

            # 获取右髋、右膝和右踝的角度
            angle_rka = math.degrees(math.atan2(right_ankle.y - right_knee.y, right_ankle.x - right_knee.x) -
                                     math.atan2(right_hip.y - right_knee.y, right_hip.x - right_knee.y))

            # 获取腰部、右髋和右肩的角度
            angle_hrs = math.degrees(math.atan2(right_shoulder.y - waist.y, right_shoulder.x - waist.x) -
                                     math.atan2(right_hip.y - waist.y, right_hip.x - waist.y))
            ##################
            # 获取鼻子、右肩和右手腕关键点的信息
            nose = results.pose_landmarks.landmark[mp_pose.PoseLandmark.NOSE]
            right_shoulder = results.pose_landmarks.landmark[mp_pose.PoseLandmark.RIGHT_SHOULDER]
            right_wrist = results.pose_landmarks.landmark[mp_pose.PoseLandmark.RIGHT_WRIST]

            # 获取鼻子、右肩和右手腕的角度
            angle_nwr = math.degrees(math.atan2(right_wrist.y - nose.y, right_wrist.x - nose.x) -
                                     math.atan2(right_shoulder.y - nose.y, right_shoulder.x - nose.x))


            # 将数据输入KNN分类器进行预测


            label = knn.predict([[angle, angle1, angle_dl, angle_dr, angle_tr, angle_tl,angle_lka,angle_hls,angle_rka,angle_hrs,angle_nwr]])



            # 在输出图片上显示角度值和动作类型
            # cv2.putText(frame, "Angle: {:.2f}".format(angle), (5, 43),
            #             cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 255), 2)
            #
            # cv2.putText(frame, "Angle_tr: {:.2f}".format(angle_tr), (5, 60),
            #             cv2.FONT_HERSHEY_SIMPLEX, 0.5, (100, 15, 255), 2)
            #
            # cv2.putText(frame, "Angle_dr: {:.2f}".format(angle_dr), (5, 80),
            #             cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 100, 255), 2)

            cv2.putText(frame, label[0], (5, 60),
                        cv2.FONT_HERSHEY_SIMPLEX, 1.1, (255, 100, 100), 2)
            # 绘制头部关键点
            nose = results.pose_landmarks.landmark[mp_pose.PoseLandmark.NOSE]
            x, y = int(nose.x * width), int(nose.y * height)
            cv2.circle(frame, (x, y), 20, (220, 200, 100), -1)

        else:
            # 如果未检测到人体，则跳过本帧处理
            cv2.putText(frame, "No body detected", (50, 50),
                        cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 0, 255), 2)

        # 将帧写入输出视频文件
        out.write(frame)

        # 显示当前帧的结果
        cv2.imshow('MediaPipe Pose Detection press q exit', frame)

        # 检测是否按下q键退出
        if cv2.waitKey(1) & 0xFF == ord('q'):
            break

# 释放资源
cap.release()
out.release()
cv2.destroyAllWindows()
基于KNN分类器的人体姿势识别算法：对视频中动作进行检测和分类

原文地址: https://www.cveoy.top/t/topic/gL0C 著作权归作者所有。请勿转载和采集!

免费AI点我,无需注册和登录