Capture video with opencv2.4.11

Autonomous Machines Jetson & Embedded Systems Jetson TX1
1

My opencv version is 2.4.11, and when I was followed by the instructions to run a test using opencv to capture the video
#include <stdio.h>
#include <opencv2/opencv.hpp>

using namespace cv;  
using namespace std;  
  
int main(int argc, char** argv)  
{  
  VideoCapture cap("nvcamerasrc ! video/x-raw(memory:NVMM), width=(int)1280, height=(int)720,format=(string)I420, framerate=(fraction)24/1 ! nvvidconv flip-method=2 ! video/x-raw, format=(string)BGRx ! videoconvert ! video/x-raw, format=(string)BGR ! appsink");  
  //VideoCapture cap("fakesrc ! videoconvert ! appsink");  
  if (!cap.isOpened())  
    {  
      cout << "Failed to open camera." << endl;  
      return -1;  
    }  
  
  for(;;)  
    {  
      Mat frame;  
      cap >> frame;  
      imshow("original", frame);  
      //waitKey(1);  
      if(waitKey(30) >= 0)  
      break;  
    }  
  return 0;  
  //cap.release();  
}

The result was “Failed to open the camera”. My camera was the onboard camera and I assume that the driver was installed because I could get the video by the instruction “nvgstcapture-1.0”. My storage is not enough to install opencv3.1 so I’m wondering whether I should uninstall the opencv 2.4. Any help is greatly appreciated.

2

Hi Aust1n,

We have some threads discussing about enable gstreamer in opencv. Currently, only opencv 3.x is verified that support gstreamer1.0. You could try to add “-DWITH_GSTREAMER=ON” in cmake and see if it works on opencv2.4.11.

Thread for 2.4.9
https://devtalk.nvidia.com/default/topic/1016103/opencv-python-and-gpu-on-tk1/

How much does your storage have?

3

Only 170MB left, I’ve tried to add that, but it still not works.

4

An alternative would be to capture video raw data with (low-level) V4L2 API and then convert the data into cv::Mat. That would work with the stock OpenCV4Tegra provided by NVIDIA.