OpenCV项目中Video Capture模块无法正常工作的求助

先看一下你提供的代码和问题描述：你说摄像头能启动但无法读取画面，且最后两行代码无效，我来帮你梳理下可能的问题点和解决办法。

首先，先贴出你当前的代码（已修正缩进，方便查看）：

import face_recognition
import cv2
import numpy as np
import csv
import os
from datetime import datetime

video_capture = cv2.VideoCapture(0)

tata_img = face_recognition.load_image_file("photos/tata.jpg")
tata_encoding = face_recognition.face_encodings(tata_img)[0]

modi_img = face_recognition.load_image_file("photos/modi.jpg")
modi_encoding = face_recognition.face_encodings(modi_img)[0]

ambani_img = face_recognition.load_image_file("photos/ambani.jpg")
ambani_encoding = face_recognition.face_encodings(ambani_img)[0]

known_face_encoding = [tata_encoding, modi_encoding, ambani_encoding]
known_face_names = ["tata", "modi", "ambani"]

students = known_face_names.copy()

face_locations = []
face_encodings = []
face_names = []
s=True

now = datetime.now()
current_date = now.strftime("%Y-%m-%d")

f = open(current_date+'.csv','w+',newline = '')
lnwriter = csv.writer(f)

while True:
    _,frame = video_capture.read()
    small_frame = cv2.resize(frame,(0,0),fx=0.25,fy=0.25)
    rgb_small_frame = small_frame[:,:,::-1]

    if s:
        face_locations = face_recognition.face_locations(rgb_small_frame)
        face_encoding = face_recognition.face_encodings(rgb_small_frame,face_locations)
        face_names = []
        for face_encoding in face_encodings:
            matches = face_recognition.compare_faces(known_face_encoding,face_encoding)
            name = ""
            face_distance = face_recognition.face_distance(known_face_encoding,face_encoding)
            best_match_index = np.argmin(face_distance)
            if matches[best_match_index]:
                name = known_face_names[best_match_index]
            face_names.append(name)
            if name in known_face_names:
                if name in students:
                    students.remove(name)
                    print(students)
                    current_time = now.strftime("%H-%M-%S")
                    lnwriter.writerow([name, current_time])

    cv2.imshow("attendence system", frame)
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

video_capture.release()
cv2.destroyAllWindows()
f.close()

# 以下是冗余的错误代码
_,frame = video_capture.read()
video_capture.release()

问题分析与解决办法 #

1. 冗余代码导致的错误

你代码最后两行_,frame = video_capture.read()和video_capture.release()是完全多余的：在循环结束后，你已经调用了video_capture.release()释放了摄像头资源，之后再调用read和release肯定会失败，直接删掉这两行即可。

2. 摄像头无法读取画面的核心问题

摄像头能启动但无法读取画面，大概率是以下几个原因：

• 摄像头索引错误：cv2.VideoCapture(0)中的0不一定是你当前使用的摄像头索引。如果是笔记本有内置摄像头+外接摄像头，或者系统默认摄像头不是0，尝试把参数换成1或者-1（自动选择可用摄像头）：

  video_capture = cv2.VideoCapture(1)  # 或者 cv2.VideoCapture(-1)
  

• 权限问题：
  • Windows系统：检查是否允许Python访问摄像头（在系统设置→隐私→摄像头中确认权限开启）。
  • Linux/Mac系统：需要确保当前用户拥有摄像头访问权限，Linux下可以执行sudo usermod -aG video $USER添加用户到video组，然后重启终端。
• OpenCV版本兼容性：虽然你安装了opencv-python和contrib，但版本可能存在冲突。建议卸载现有版本后重新安装稳定版：

  pip uninstall opencv-python opencv-contrib-python
  pip install opencv-python==4.5.5.62 opencv-contrib-python==4.5.5.62
  

3. 代码逻辑的其他小问题

除了摄像头的问题，你的代码还有几个逻辑小错误，会影响考勤功能：

• 时间不更新：now = datetime.now()是在循环外定义的，只会获取程序启动时的时间，应该把current_time = datetime.now().strftime("%H-%M-%S")放在循环内，每次识别到人脸时获取当前时间。
• 缩进错误：原来的if name in known_face_names:这段代码缩进不正确，应该放在for循环内部，否则逻辑会混乱（已经在上面的修正代码中调整）。

修正后的完整代码 #

这里给你整理了修正后的代码，解决了上述问题：

import face_recognition
import cv2
import numpy as np
import csv
import os
from datetime import datetime

# 尝试更换摄像头索引，0不行换1或-1
video_capture = cv2.VideoCapture(0)

tata_img = face_recognition.load_image_file("photos/tata.jpg")
tata_encoding = face_recognition.face_encodings(tata_img)[0]

modi_img = face_recognition.load_image_file("photos/modi.jpg")
modi_encoding = face_recognition.face_encodings(modi_img)[0]

ambani_img = face_recognition.load_image_file("photos/ambani.jpg")
ambani_encoding = face_recognition.face_encodings(ambani_img)[0]

known_face_encoding = [tata_encoding, modi_encoding, ambani_encoding]
known_face_names = ["tata", "modi", "ambani"]

students = known_face_names.copy()

face_locations = []
face_encodings = []
face_names = []
s=True

current_date = datetime.now().strftime("%Y-%m-%d")

f = open(current_date+'.csv','w+',newline = '')
lnwriter = csv.writer(f)

while True:
    # 检查read是否成功
    ret, frame = video_capture.read()
    if not ret:
        print("无法读取摄像头画面，请检查摄像头连接或索引")
        break
        
    small_frame = cv2.resize(frame,(0,0),fx=0.25,fy=0.25)
    rgb_small_frame = small_frame[:,:,::-1]

    if s:
        face_locations = face_recognition.face_locations(rgb_small_frame)
        face_encodings = face_recognition.face_encodings(rgb_small_frame,face_locations)
        face_names = []
        for face_encoding in face_encodings:
            matches = face_recognition.compare_faces(known_face_encoding,face_encoding)
            name = ""
            face_distance = face_recognition.face_distance(known_face_encoding,face_encoding)
            best_match_index = np.argmin(face_distance)
            if matches[best_match_index]:
                name = known_face_names[best_match_index]
            face_names.append(name)
            if name in known_face_names and name in students:
                students.remove(name)
                print(f"已签到：{name}，剩余未签到：{students}")
                # 每次获取当前时间
                current_time = datetime.now().strftime("%H-%M-%S")
                lnwriter.writerow([name, current_time])

    cv2.imshow("attendence system", frame)
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

video_capture.release()
cv2.destroyAllWindows()
f.close()

可以先尝试更换摄像头索引，然后运行修正后的代码，应该能解决摄像头无法读取的问题。如果还是不行，再检查权限和OpenCV版本。

备注：内容来源于stack exchange，提问作者Darshan