Ubuntu 18.04（TX2平台）下Webcam无法打开及显示的问题求助

Hey there! Let's tackle this webcam issue step by step—since you confirmed it works with nvgstcapture-1.0, we know the hardware and low-level drivers are in good shape. The problem is likely tied to GStreamer configuration, plugin compatibility, or permission settings. Here's what to try:

1. Verify Webcam's Supported Formats First #

Before tweaking pipelines, confirm your webcam actually supports the MJPG parameters you're using. Run this command to list all supported formats and resolutions:

v4l2-ctl -d /dev/video1 --list-formats-ext

Check if there's an exact match for MJPG with 640x480 at 30fps. If not, adjust your pipeline to use a resolution/framerate the camera actually supports.

2. Fix GStreamer Pipeline Issues #

Your initial pipeline uses hardware-accelerated elements, but let's start simple to isolate the problem:

Test a Basic (Non-Hardware-Accelerated) Pipeline #

Run this to rule out hardware decoding conflicts:

gst-launch-1.0 v4l2src device=/dev/video1 ! image/jpeg,format=MJPG,width=640,height=480,framerate=30/1 ! jpegdec ! autovideosink

If this shows video, the issue is with the hardware-accelerated part of your original pipeline. Try this adjusted hardware pipeline (using nveglglessink which is better optimized for TX2):

gst-launch-1.0 v4l2src device=/dev/video1 ! image/jpeg,format=MJPG,width=640,height=480,framerate=30/1 ! nvjpegdec ! nvvidconv ! nveglglessink

Fix the "Internal Data Stream Error" #

That error often happens if the pipeline can't keep up with the camera's output. Adding a queue element can help buffer data:

gst-launch-1.0 v4l2src device=/dev/video1 ! queue ! image/jpeg,format=MJPG,width=640,height=480,framerate=30/1 ! nvjpegdec ! queue ! nvvidconv ! nveglglessink

3. Ensure GStreamer Plugins Are Complete #

You installed some plugins, but missing ones might be causing issues. Install the full set of common plugins:

sudo apt-get update && sudo apt-get install gstreamer1.0-plugins-good gstreamer1.0-plugins-ugly gstreamer1.0-tools

Also, reinstall the bad plugins package to fix the earlier videoparsers error (don't ignore it—it might be needed for MJPG parsing):

sudo apt-get install --reinstall gstreamer1.0-plugins-bad

4. Fix Cheese's Grayed-Out Device #

Cheese might be missing permissions or using the wrong backend:

• Add your user to the video group to ensure access to the webcam device:

  sudo usermod -aG video $USER
  

  Log out and back in for this change to take effect.
• If that doesn't work, try launching Cheese with explicit GStreamer debug output to see what's failing:

  GST_DEBUG=3 cheese
  

  Look for errors related to v4l2src or device access in the output.

5. Fix OpenCV & MATLAB Issues #

OpenCV #

Make sure your OpenCV was compiled with GStreamer support (check with cv::getBuildInformation()). Then use a GStreamer pipeline directly in your code:

#include <opencv2/opencv.hpp>

int main() {
    std::string pipeline = "v4l2src device=/dev/video1 ! image/jpeg,format=MJPG,width=640,height=480,framerate=30/1 ! nvjpegdec ! nvvidconv ! video/x-raw,format=BGRx ! videoconvert ! video/x-raw,format=BGR ! appsink";
    cv::VideoCapture cap(pipeline, cv::CAP_GSTREAMER);
    
    if (!cap.isOpened()) {
        std::cerr << "Failed to open webcam!" << std::endl;
        return -1;
    }

    cv::Mat frame;
    while (cap.read(frame)) {
        cv::imshow("Webcam Output", frame);
        if (cv::waitKey(1) == 'q') break;
    }

    cap.release();
    cv::destroyAllWindows();
    return 0;
}

MATLAB #

MATLAB's webcam support might be using an outdated GStreamer configuration. Try specifying the GStreamer pipeline manually in MATLAB:

pipeline = 'v4l2src device=/dev/video1 ! image/jpeg,format=MJPG,width=640,height=480,framerate=30/1 ! nvjpegdec ! nvvidconv ! video/x-raw,format=BGRx ! videoconvert ! video/x-raw,format=BGR ! appsink';
cam = cv.VideoCapture(pipeline, cv.CAP_GSTREAMER);
while isOpened(cam)
    [ret, frame] = read(cam);
    if ~ret, break; end
    imshow(frame);
    if waitKey(1) == 'q', break; end
end
release(cam);
close all;

Note: This requires the Computer Vision Toolbox with GStreamer support enabled.

Start with step 1 and work your way through—most likely the pipeline adjustment or plugin installation will fix the issue. Let me know if any of these steps work or if you hit new errors!

内容的提问来源于stack exchange，提问作者slow_one