Visual-based automatic video feed selection for a digital video production system
A video production device is deployed to produce a video production stream of an event occurring within an environment that includes a plurality of different video capture devices capturing respective video input streams of the event. The video production device is programmed and operated to: receive a plurality of video input streams from the video capture devices; automatically select one of the video input streams as a current video output stream, in accordance with a visual activity based switching technique that identifies which of the plurality of video input streams is associated with a detected speaker; and provide the selected video input stream as the current video output stream.
1. An automated method of controlling a video production device that produces a video production stream of an event from a plurality of video input streams that are captured by a plurality of different video capture devices, the method comprising:
receiving the plurality of video input streams from the plurality of different video capture devices;
automatically selecting one of the plurality of video input streams as a current video output stream, in accordance with a visual activity based switching technique that identifies which of the plurality of video input streams is associated with a detected speaker; and
providing the selected video input stream as the current video output stream;
wherein the visual activity based switching technique identifies and analyzes lip and mouth positions captured in the plurality of video input streams, and calculates a distance ratio from the lip and mouth positions to identify which of the plurality of video input streams is associated with a detected speaker; and
wherein the visual activity based switching technique identifies which of the plurality of video input streams is associated with a detected speaker based on a variation in distance ratio given by the expression
(
V
T
-
V
B
)
(
H
L
-
H
R
)
,
wherein H L and H R denote left and right corner positions of the mouth of the detected speaker, and wherein V T and V B denote top and bottom lip positions of the detected speaker.
2. The method of claim 1 , further comprising the step of producing the current video output stream for storage or live streaming.
3. The method of claim 1 , wherein the visual activity based switching technique identifies the detected speaker when sample-to-sample variation in the distance ratio exceeds a threshold value.
4. The method of claim 1 , wherein the visual activity based switching technique determines which of the plurality of video input streams corresponds to a highest calculated distance ratio, and automatically selects the determined video input stream as the current video output stream.
5. A video production device for producing a video production stream of an event occurring within an environment that includes a plurality of different video capture devices capturing respective video input streams of the event, the video production device comprising:
a processor hardware device; and
a memory storage device to store processor-executable instructions that are configurable to be executed by the processor hardware device to perform a method comprising the steps of:
receiving a plurality of video input streams from the plurality of different video capture devices;
automatically selecting one of the plurality of video input streams as a current video output stream, in accordance with a visual activity based switching technique that identifies which of the plurality of video input streams is associated with a detected speaker; and
providing the selected video input stream as the current video output stream;
wherein the visual activity based switching technique identifies and analyzes lip and mouth positions captured in the plurality of video input streams, and calculates a distance ratio from the lip and mouth positions to identify which of the plurality of video input streams is associated with a detected speaker; and
wherein the visual activity based switching technique identifies which of the plurality of video input streams is associated with a detected speaker based on a variation in distance ratio given by the expression
(
V
T
-
V
B
)
(
H
L
-
H
R
)
,
wherein H L and H R denote left and right corner positions of the mouth of the detected speaker, and wherein VT and VB denote top and bottom lip positions of the detected speaker.
6. The video production device of claim 5 , wherein the method performed by the processor hardware device further comprises the step of producing the current video output stream for storage or live streaming.
7. The video production device of claim 5 , wherein the visual activity based switching technique identifies the detected speaker when sample-to-sample variation in the distance ratio exceeds a threshold value.
8. The video production device of claim 5 , wherein the visual activity based switching technique determines which of the plurality of video input streams corresponds to a highest calculated distance ratio, and automatically selects the determined video input stream as the current video output stream.
9. A video production system comprising:
a plurality of video capture devices; and
a video production device to wirelessly communicate with the plurality of video capture devices within a wireless network established by the video production device, the video production device comprising a processor hardware device and a memory storage device to store processor-executable instructions that are configurable to be executed by the processor hardware device to perform a method comprising the steps of:
receiving a plurality of video input streams from the plurality of different video capture devices;
automatically selecting one of the plurality of video input streams as a current video output stream, in accordance with a visual activity based switching technique that identifies which of the plurality of video input streams is associated with a detected speaker; and
providing the selected video input stream as the current video output stream;
wherein the visual activity based switching technique identifies and analyzes lip and mouth positions captured in the plurality of video input streams, and calculates a distance ratio from the lip and mouth positions to identify which of the plurality of video input streams is associated with a detected speaker; and
wherein the visual activity based switching technique identifies which of the plurality of video input streams is associated with a detected speaker based on a variation in distance ratio given by the expression
(
V
T
-
V
B
)
(
H
L
-
H
R
)
,
wherein H L and H R denote left and right corner positions of the mouth of the detected speaker, and wherein V T and V B denote top and bottom lip positions of the detected speaker.
10. The video production system of claim 9 , wherein the visual activity based switching technique identifies the detected speaker when sample-to-sample variation in the distance ratio exceeds a threshold value.
11. The video production system of claim 9 , wherein the visual activity based switching technique determines which of the plurality of video input streams corresponds to a highest calculated distance ratio, and automatically selects the determined video input stream as the current video output stream.