WebCamera | Camera controls for the Web | Camera library

There are several methods to solve the problem.

IMPORTANT: I do not recommend using the WebcamJS package because it is an unmaintained package and there are other viable alternatives. You can fork the package and fix any errors but that would be more work.

Option #1 (Maybe the best choice for you)

My advice is to create a custom field that extends "Avatar" or "File" and add the "Capture from webcam" option. You can use one of these packages that I link below (more up-to-date than WebcamJS) quickly and easily.

Option #2 (Easy 'n shared!)

Another possible, even simpler approach is to create a standalone custom "Webcam" field. You will still need the Nova Field Dependency Container package but your custom field could share it with the community in a standalone way.

How to create a custom field in laravel nova

Alternative packages to WebcamJS:

• Multicorder
• VueWebCam
• EasyVueCamera
• VueMediaRecorder

Choose the package that is most useful for you! I hope to be proved helpful!

Assuming you actually need to use OpenCV for the video, you have to add the audio directly to FFmpeg as Gyan commented, because OpenCV does not support audio.

-re argument is probably required for live streaming.

For testing, I modified the RTMP URL from YouTube to localhost.
FFplay sub-process is used for capturing the stream (for testing).

Complete code sample:

Have you tried this?

You could try and set the time to 12:00:01am at 11:59:59pm. I'm not sure if that would fix it. A while loop could work.

SSHFS allows you to mount a remote directory using SSH.

1. Make target directory on remote host. In this example it is named remotedir.
2. Make mount directory on local host. In this example it is named mountpoint.
3. Run sshfs: sshfs username@"$vps_ip":remotedir mountpoint
4. Run ffmpeg: ffmpeg -f v4l2 -i /dev/video0 -c:v libx264 -profile:v baseline -vf format=yuv420p mountpoint/output.flv
5. To unmount mountpoint run fusermount3 -u mountpoint (Linux) or umount mountpoint (macOS).

As for your ffmpeg command do not use -re with live inputs, and there is no need for -strict -2.

problem was the tensorflow version , do the following: uninstall current version

Ok, I found the solution. It's so good that we can't pass HTML elements as parameters to web worker, but ofc we can play with canvas context as below:

Ability to save/load properties for DirectShow objects representing webcameras is optional. Having said that, "Query... failed." and "I/O error" is not exactly an error. It is non-fatal error, that is. There is no problem at the camera side.

Webcamera driver implementation might have its own internal persistence, such as, for example, when you change some property like contrast setting it might be stored and be used further for all application consuming the camera.

To summarize our discussion into an answer that can be accepted:

1. Remove cv2.imshow to get rid of the second window
2. Use img = Image.fromarray(cv2.cvtColor(frame, cv2.COLOR_BGR2RGBA)) to display the frame with the detected face with the right colors.