python - OpenCV 背景减法学习率不能改变

Question

我希望用 50 帧训练背景区域，并使用此预训练模型进行背景减法。模型在训练后停止更新。

这是我的代码

import cv2
print "This program is for background subtraction with pre-trained model\n"

Training_Floder = "/Users/yuyang/Desktop/img1/"
Start_Frame_Num = 1
End_Frame_Num = 51

cv2.namedWindow("BG_IMAGE")

fgbg = cv2.createBackgroundSubtractorMOG2(50, 16, False)
font = cv2.FONT_HERSHEY_SIMPLEX



for index in range(Start_Frame_Num, End_Frame_Num):
    Img_File_Name = Training_Floder + str(index) + ".jpg"
    Img = cv2.imread(Img_File_Name)
    fgmask = fgbg.apply(Img, -1)
    BG_IMG = fgbg.getBackgroundImage()
    #######
    cv2.putText(BG_IMG,str(index),(10,500), font, 1,(255,255,255),2)
    cv2.imshow("BG_IMAGE", BG_IMG)
    #######
    cv2.waitKey(0)

Testing_Floder = "/Users/yuyang/Desktop/New/"
Test_Start = 1
Test_End = 100

for index in range(Test_Start, Test_End):
    Img_File_Name = Testing_Floder + str(index) + ".jpg"
    Img = cv2.imread(Img_File_Name)
    fgmask1 = fgbg.apply(Img, 0)
    BG_IMG1 = fgbg.getBackgroundImage()
    cv2.putText(BG_IMG1,str(index),(10,500), font, 1,(255,255,255),2)
    cv2.imshow("BG_IMAGE", BG_IMG1)
    cv2.waitKey(0)

根据评论

学习率参数在函数“apply()”中。

@param learningRate 
The value between 0 and 1 that indicates how fast the background 
model is learnt. Negative parameter value makes the algorithm to 
use some automatically chosen learning rate. 0 means that the 
background model is not updated at all, 1 means that the background 
model is completely reinitialized from the last frame.

CV_WRAP virtual void apply(InputArray image, OutputArray fgmask, double learningRate=-1) = 0;"

但是，我在这里尝试了几种学习率:

fgmask = fgbg.apply(Img, -1) or
fgmask = fgbg.apply(Img, 0) or
fgmask = fgbg.apply(Img, 1) or
fgmask = fgbg.apply(Img, 0.00001)

训练背景结果不会改变。
这意味着我不能在测试时保持预训练模型不变!

我的代码有什么问题吗？
有没有办法改变学习率？

以下是一些结果

Background subtraction result of Testing image #1

Background subtraction result of Testing image #40

从上面的结果可以看出，虽然我将学习率设置为 0，但训练后的背景图像在测试时会发生变化。

fgmask1 = fgbg.apply(Img, 0)

Answer 1

所以使用python实现的正确方法是

fgbg = cv2.createBackgroundSubtractorMOG2(50, 16, False)
fgbg.apply(input, output, learning_rate)

与在 c++ 实现中完全一样。
学习率必须是第三个参数。