使用 mediapipe 库时是否可以不显示视频源？答案

【问题标题】：is it possible to not display the video feed when using the mediapipe library?使用 mediapipe 库时是否可以不显示视频源？
【发布时间】：2021-10-19 20:56:22
【问题描述】：

我试图只显示地标和 2d 插图，而不显示它们背后的视频源，但无济于事。这似乎我应该有一个非常简单的答案，但我迷路了。有人知道如何去做吗？

    #dependencies
import cv2
import mediapipe as mp
import numpy as np
mp_drawing = mp.solutions.drawing_utils
mp_holistic = mp.solutions.holistic
cap = cv2.VideoCapture(0)
#mediapipe instance
with mp_holistic.Holistic(min_detection_confidence=0.5, min_tracking_confidence=0.5) as holistic:

    while cap.isOpened():
        ret, frame = cap.read()
        frame.flags.writeable = False
        
        #make detection
        results = holistic.process(frame)

        #render detections
        mp_drawing.draw_landmarks(frame, results.face_landmarks, mp_holistic.FACE_CONNECTIONS, 
                                 mp_drawing.DrawingSpec(color=(256, 256, 256), thickness=1, circle_radius=1),
                                 mp_drawing.DrawingSpec(color=(256, 256, 256), thickness=1, circle_radius=1)
                                 )
        mp_drawing.draw_landmarks(frame, results.pose_landmarks, mp_holistic.POSE_CONNECTIONS,
                                 mp_drawing.DrawingSpec(color=(256, 256, 256), thickness=1, circle_radius=1),
                                 mp_drawing.DrawingSpec(color=(256, 256, 256), thickness=1, circle_radius=1)
                                 )
        mp_drawing.draw_landmarks(frame, results.right_hand_landmarks, mp_holistic.HAND_CONNECTIONS, 
                                 mp_drawing.DrawingSpec(color=(256, 256, 256), thickness=1, circle_radius=1),
                                 mp_drawing.DrawingSpec(color=(256, 256, 256), thickness=1, circle_radius=1)
                                 )
        mp_drawing.draw_landmarks(frame, results.left_hand_landmarks, mp_holistic.HAND_CONNECTIONS, 
                                 mp_drawing.DrawingSpec(color=(256, 256, 256), thickness=1, circle_radius=1),
                                 mp_drawing.DrawingSpec(color=(256, 256, 256), thickness=1, circle_radius=1)
                                 )
        
        cv2.imshow('frame', frame)
        if cv2.waitKey(10) & 0xFF == ord('q'):
            break
        
cap.release()
cv2.destroyAllWindows()

【问题讨论】：

嗨@aidan.goetzinger :) 欢迎来到stackoverflow。我认为您应该删除 cv2.imshow('frame', frame) 行
或者你可以制作一个空图像，绘制地标并显示它而不是框架:)

标签： python opencv artificial-intelligence mediapipe

【解决方案1】：

加载一张背景为空白的图片（将其调整为相机输入大小，例如 1280x720 并为其命名，例如 blank_bg

blank_bg = cv2.imread("image")

在 mp_draw 函数中，将“frame”替换为“blank_bg”。

mp_drawing.draw_landmarks(blank_bg, results.face_landmarks, mp_holistic.FACE_CONNECTIONS, 
                         mp_drawing.DrawingSpec(color=(256, 256, 256), thickness=1, circle_radius=1),
                         mp_drawing.DrawingSpec(color=(256, 256, 256), thickness=1, circle_radius=1)
                         )

最后...

cv.imshow("frame",blank_bg)

【讨论】：