Cannot write video frame using OpenCV VideoWriter

Autonomous Machines Jetson & Embedded Systems Jetson TX1
1

Hi all,

I used OpenCV4Tegra(Version 2.4.13) and Python to record the video. However, I always get the blank result video. Does anyone face the same problem like me?

A piece of code below:

fourcc = cv2.cv.CV_FOURCC(‘X’, ‘V’, ‘I’, ‘D’)
videoRecorder = cv2.VideoWriter(‘pcs_video.avi’,fourcc,20.0,(640,480))
…
videoRecorder.write(frame)

2

HI,

Please try my code

import sys
import cv2

def read_cam():
    cap = cv2.VideoCapture(1)
    # Define the codec and create VideoWriter object
    fourcc = cv2.cv.CV_FOURCC(*'XVID')
    out = cv2.VideoWriter('output.avi',fourcc, 20.0, (640,480))
    if cap.isOpened():
        while True:
            ret_val, img = cap.read();
            out.write(img)
            #cv2.imshow('demo',img) 
            #cv2.waitKey(10)

    else:
     print "camera open failed"

    out.release
    cv2.destroyAllWindows()


if __name__ == '__main__':
    read_cam()
3

Hi,

Thank you for your sample code but I’ve not yet done with this problem.
My purpose’s to record the color and depth frame from Kinect Xbox One using libfreenect2 & OpenCV4Tegra. However, I always get the blank video.

fourcc = cv2.cv.CV_FOURCC(*‘XVID’)
videoRecorder = cv2.VideoWriter(‘video_2x.avi’,fourcc,20.0,(int(1920 / 3), int(1080 / 3))) ## Resize color frame (1920,1080)

…

rgbFrame = cv2.resize(color.asarray(),(int(1920 / 3), int(1080 / 3)))
videoRecorder.write(rgbFrame)

Thank you.

4

Hi Kata,

So the frames from Kinect are okay for usecase like preview? The problem only happens when recording?

5

Hi WayneWWWW,

Thank you for your prompt rely.

So the frames from Kinect are okay for usecase like preview?The problem only happens when recording?
Yes, I can view the frame correctly by displaying the frame. The problem is come from recording.

6

To align our environment, you are using rel-28.1 from Jetpack with opencv4tegra, right? Do you install full package?

7

Could you try usb camera first? Because my code was using a web cam for testing.

8

Hi WayneWWW,

Thank you so much for your strong support!

To align our environment, you are using rel-28.1 from Jetpack with opencv4tegra, right? Do you install full >package?
Yes, it’s right. I installed full package.

Could you try usb camera first? Because my code was using a web cam for testing.
Yeath, your code’s been worked perfectly. That means the encoder also works correctly.

Any suggestion for my problem?

9

Hi,

Do you read depth data into cv::Mat?
If yes, which format do you use?

Usually, we use CV_8UC3 for camera input, which stands for [0-255] color image.
If you not in CV_8UC3, there is possibly an auto-truncate when encoding the video. This truncate may cause some information loss.

Thanks.

10

This might be very late for you but might help someone else. I also went through the same problem recently.
Found out that the video dimensions of the output file should match the dimension of the frame
you are trying to write.

out = cv2.VideoWriter(‘output.avi’,fourcc, 20.0, (W,H))

where W, H are width and height of video frames respectively.

If you are reading the input frame with VideoCapture object to read frames from another video stream as
cap = cv2.VideoCapture(1)

use
W = cap.get(3)
H = cap.get(4)

If you are reading the frame from another source or method,
use

W = input_frame.shape[0]
H = input_frame.shape[1]

assuming input_frame is an numpy array.

Hope this helps.

11

Nice! Thanks for sharing!