Video-Based Active Communicator Detection for Noisy Conferences
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing conferencing systems inaccurately identify active communicators based on audio levels, leading to errors such as misidentifying participants due to background noise or failing to recognize sign language, and do not account for non-verbal cues.
Innovation Solution
An electronic conferencing system identifies active communicators through video analysis, using deep neural networks to recognize body language, sign language, and other non-verbal cues, and graphically distinguishes them using GUI enhancements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If active communicator identification is based on audio level detection, then the system can identify speakers in real-time, but background noise causes misidentification of active communicators
Solution Approach 1:
The patent segments the identification process into multiple independent analysis channels: audio level detection, video frame analysis, and motion detection. Each channel processes different aspects of participant behavior independently, then combines results to make a comprehensive determination. This segmentation allows the system to overcome the limitations of audio-only detection by incorporating visual cues that are independent of background noise.
Solution Approach 2:
The patent introduces video frames as an intermediary element between the participant and the identification system. By analyzing changes in video frames between consecutive time points, the system obtains an independent verification signal that mediates the identification process. This intermediary visual information helps distinguish genuine speaking activity from background noise artifacts.
2Device complexity
If audio level detection is used to identify active communicators, then the system operates with simple processing, but it fails to recognize participants using sign language
Solution Approach 1:
The patent makes the identification system universal by enabling it to handle multiple communication modes through a single integrated framework. The video analysis component serves multiple functions: it detects speech-related mouth movements for traditional speakers, recognizes sign language gestures for deaf participants, and identifies hand-raising actions for those seeking to speak. This multi-functionality allows the system to adapt to diverse communication needs without requiring separate specialized systems.
Solution Approach 2:
The patent transitions from one-dimensional audio-based identification to two-dimensional video-based identification, adding the visual dimension to the recognition process. By analyzing spatial-temporal patterns in video frames, the system captures gestures, facial expressions, and body movements that occur in the visual domain. This dimensional expansion enables recognition of non-audible communication methods such as sign language while maintaining compatibility with traditional speech detection.
3Measurement precision
If the system processes video data to identify gestures and movements, then communication clarity improves, but processing complexity and computational requirements increase
Solution Approach 1:
The patent applies partial action by selectively processing only the most informative aspects of video data. Rather than analyzing every pixel and movement in detail, the system focuses on detecting specific types of changes between consecutive frames that are most relevant to communication intent. This selective processing reduces computational complexity while maintaining high accuracy for identifying speaking turns and gestures.
Solution Approach 2:
The system performs preliminary processing of video data by pre-defining regions of interest and expected gesture patterns before full analysis. By establishing baseline expectations for communication-related movements and pre-processing video frames to highlight relevant features, the system reduces the computational burden during real-time identification. This preliminary preparation enables faster, more efficient processing during actual conferencing operations.
Data Source
AI summary
A method includes receiving a plurality of video data associated with a plurality of users in an electronic conference. The method further includes processing the plurality of video data to determine whether an action of a user matches an active communicator action. In response to determining that the action of the user matches the active communicator action, the method further includes determining that the user intends to actively communicate audio or video data in the electronic conference. Moreover, the method includes identifying the user as an active communicator within the electronic conference.