如何在MediaPipe与OpenCV中区分左右手同编号关键点？

问题根源 #

你的代码里，遍历每只手的关键点时，始终调用results.multi_handedness[0]获取标签——这意味着不管当前处理的是哪只手，都只会用第一只检测到的手的左右标签，导致两只手的同编号关键点被标记成同一个颜色。

修正后的完整代码 #

只需要在遍历手的时候，同步获取当前手的索引，对应到multi_handedness的对应项就能解决：

import cv2
import mediapipe as mp
import time

cap = cv2.VideoCapture(0)
mpHands = mp.solutions.hands
hands = mpHands.Hands(max_num_hands=2)
mpDraw = mp.solutions.drawing_utils

while True:    
    success, img = cap.read()
    if not success:
        break
    img = cv2.flip(img, 1) 
    # time.sleep(0.1)  # 该语句会降低帧率，建议按需保留
    imgRGB = cv2.cvtColor(img,cv2.COLOR_BGR2RGB)

    results = hands.process(imgRGB)
    
    if results.multi_hand_landmarks:
        # 遍历手时同时获取索引，对应到multi_handedness的对应项
        for hand_idx, handLms in enumerate(results.multi_hand_landmarks):        
            # 获取当前手的左右标签
            lbl = results.multi_handedness[hand_idx].classification[0].label
            
            for id, lm in enumerate(handLms.landmark):
                h,w,c = img.shape    
                cx, cy = int(lm.x*w), int(lm.y*h)
                
                # 标记关键点#8
                if id == 8:
                    if lbl == 'Right':
                        cv2.circle(img, (cx,cy), 8, (255,0,0), cv2.FILLED)
                    else:
                        cv2.circle(img, (cx,cy), 8, (255,255,0), cv2.FILLED)

            mpDraw.draw_landmarks(img, handLms, mpHands.HAND_CONNECTIONS,
                mpDraw.DrawingSpec(color=(121,0,255)),
                mpDraw.DrawingSpec(color=(0,255,0)))
    
    cv2.imshow("image", img)
    c = cv2.waitKey(7) % 0x100
    if c == 27:
        break

cap.release()
cv2.destroyAllWindows()

提取左右手全部关键点的方法 #

如果需要单独提取并存储左右手所有40个关键点的坐标，可以用字典分类存储：

# 在循环外初始化存储容器
left_hand_coords = {}
right_hand_coords = {}

# 在处理results的循环内
if results.multi_hand_landmarks:
    for hand_idx, handLms in enumerate(results.multi_hand_landmarks):
        lbl = results.multi_handedness[hand_idx].classification[0].label
        for id, lm in enumerate(handLms.landmark):
            h,w,c = img.shape    
            cx, cy = int(lm.x*w), int(lm.y*h)
            # 按标签存入对应字典，键为关键点ID
            if lbl == 'Right':
                right_hand_coords[id] = (cx, cy)
            else:
                left_hand_coords[id] = (cx, cy)

# 之后可以直接通过ID获取对应坐标
# 比如获取右手关键点#8：right_hand_coords.get(8)
# 获取左手关键点#8：left_hand_coords.get(8)

内容的提问来源于stack exchange，提问作者Leo