有没有一种方法可以让我在视频的中间进行我的车辆检测模型。

Question

我只想在视频中心对我的车辆检测模型进行推断。就像你在这个图片里看到的。红色区域只是我希望我的模型运行的地方。我想知道是否有办法让我这么做。若要指定模型工作的区域，请执行以下操作。

Answer 1

这是我用在作物上的代码。我只需要用我的车辆检测模型来实现它，在我完成之后我就会发布它。

# Import packages
import cv2
import numpy as np

# Open the video
cap = cv2.VideoCapture('test.mp4')

# Initialize frame counter
cnt = 0

# Some characteristics from the original video
w_frame, h_frame = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH)), 
int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
fps, frames = cap.get(cv2.CAP_PROP_FPS), cap.get(cv2.CAP_PROP_FRAME_COUNT)

# Here you can define your croping values
x,y,h,w = 0,0,600,1000

# output
fourcc = cv2.VideoWriter_fourcc(*'MP4V')

#fourcc = cv2.cv.CV_FOURCC(*'XVID')

out = cv2.VideoWriter('result.mp4', fourcc, fps, (w, h))


# Now we start
while(cap.isOpened()):
ret, frame = cap.read()

    cnt += 1 # Counting frames

# Avoid problems when video finish
if ret==True:
    # Croping the frame
    crop_frame = frame[y:y+h, x:x+w]

    # Percentage
    xx = cnt *100/frames
    print(int(xx),'%')

    # Saving from the desired frames
    #if 15 <= cnt <= 90:
    #    out.write(crop_frame)

    # I see the answer now. Here you save all the video
    out.write(crop_frame)

    # Just to see the video in real time          
    cv2.imshow('frame',frame)
    cv2.imshow('croped',crop_frame)

    if cv2.waitKey(1) & 0xFF == ord('q'):
        break
else:
    break


cap.release()
out.release()  
cv2.destroyAllWindows()