如何修复cvzone手部检测中ValueError: too many values to unpack错误？

我编写了结合cvzone手部检测与MCQ展示的Python程序，期望实现食指指尖（lmList[8]）与中指指尖（lmList[12]）的距离检测，但运行时触发以下错误：

Traceback (most recent call last):
  File "e:\CODE\.vscode\OpenCV\main.py", line 57, in <module>
    length, info = detector.findDistance(lmList[0],lmList[12])
  File "C:\Users\User1\AppData\Local\Programs\Python\Python310\lib\site-packages\cvzone\HandTrackingModule.py", line 143, in findDistance
    x1, y1 = p1
ValueError: too many values to unpack (expected 2)

我尝试更换不同版本的mediapipe，问题仍未解决。完整代码如下：

import cv2 as cv
import csv
from cvzone.HandTrackingModule import HandDetector
import cvzone as cvz

# Declaring the cap variable to start Camera activity

cap = cv.VideoCapture(0)
cap.set(3, 1280)
cap.set(4, 720)

detector = HandDetector(minTrackCon=0.8)  


# Class to access the elements of Que.csv file via indexing

class Questions:
    def __init__(self, data):
        self.question = data[0]
        self.choice1 = data[1]
        self.choice2 = data[2]
        self.choice3 = data[3]
        self.choice4 = data[4]
        self.answer = int(data[5])
        self.userAnswer = None

# importing CSV file

getFile = 'Que.csv'
with open(getFile, newline='\n') as file:
    reader = csv.reader(file)
    datafile = list(reader)[1:]

# Creating Object for Each

mcqList = []
for que in datafile:
    mcqList.append(Questions(que))

print(len(mcqList))

queNumber = 0
queTotal = len(datafile)

while True:
    ret, frame = cap.read()
    # To Flip the camera view to the original side
    hands, frame = detector.findHands(frame, flipType=True)
    mcq = mcqList[0]

    frame, box = cvz.putTextRect(frame, mcq.question, [50, 50], 2, 2, offset=15, border=4)
    frame, box1 = cvz.putTextRect(frame, mcq.choice1, [50, 115], 1, 2, offset=15, border=4)
    frame, box2 = cvz.putTextRect(frame, mcq.choice2, [150, 115], 1, 2, offset=15, border=4)
    frame, box3 = cvz.putTextRect(frame, mcq.choice3, [50, 165], 1, 2, offset=15, border=4)
    frame, box4 = cvz.putTextRect(frame, mcq.choice4, [150, 165], 1, 2, offset=15, border=4)

    # Checking for a Hand
    if hands:
        lmList = hands[0]['lmList']
        length, info, frame = detector.findDistance(lmList[8],lmList[12],frame)
        print(length)

    cv.imshow('frame', frame)
    if cv.waitKey(1) & 0xFF == ord('q'):
        break

cap.release()
cv.destroyAllWindows()

错误原因 #

lmList中的每个关键点包含x、y、z三个坐标值，而findDistance函数内部尝试将传入的点解包为x和y两个值，直接传入完整的lmList[8]或lmList[12]会因为多了z轴值导致解包失败。

修复步骤 #

调用findDistance时，仅传入关键点的x和y坐标，忽略z轴数据，修改手部检测部分的代码：

# Checking for a Hand
if hands:
    lmList = hands[0]['lmList']
    # 提取每个指尖的x、y坐标（忽略z值）
    length, info, frame = detector.findDistance(lmList[8][:2], lmList[12][:2], frame)
    print(length)

额外说明 #

• 错误日志中写的是detector.findDistance(lmList[0],lmList[12])，但实际代码用的是lmList[8]，确保你取的是食指（索引8）和中指（索引12）的指尖关键点，避免笔误。
• findDistance在传入frame参数时会返回三个值：length（距离）、info（包含中点坐标等信息的字典）、frame（绘制了连线的帧），确保接收变量数量匹配。

内容的提问来源于stack exchange，提问作者Roshan