Video Encoder Parameter Tuning for Event Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video analytics systems face challenges in processing massive amounts of video data from cameras, as only a small subset of frames is critical for real-time decisions, and current methods are inefficient in selecting these frames without impacting object recognition performance.
Innovation Solution
A data-driven event detection method that tunes video encoding parameters, such as GOP size and scenecut threshold, to select a subset of video frames for machine learning-based object recognition, allowing for efficient processing and maintaining recognition performance by generating I-frames only when events occur.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If video analytics systems process all video frames to ensure accurate event detection, then detection accuracy is improved, but processing cost and time consumption increase prohibitively
Solution Approach 1:
The patent segments video frames into two categories: I-frames (independent frames containing complete image data) and P-frames (dependent frames containing only differences from previous frames). The system processes only I-frames for event detection while using P-frames for motion detection and frame reconstruction, significantly reducing the number of frames requiring full processing while maintaining detection accuracy.
Solution Approach 2:
The patent extracts and processes only the critical subset of video frames (I-frames) that contain complete visual information necessary for event detection. By extracting these key frames and using them as reference points, the system avoids processing the entire video stream while maintaining detection performance.
2Measurement precision
If video encoders generate I-frames frequently to improve event detection, then detection accuracy improves, but bandwidth consumption and storage requirements increase
Solution Approach 1:
The patent dynamically adjusts the I-frame generation frequency based on scene complexity and motion detection results. During periods of low activity, the system reduces I-frame generation to minimize bandwidth consumption. During periods of high activity or detected events, the system increases I-frame generation to ensure accurate event capture, optimizing the balance between detection accuracy and bandwidth usage.
Solution Approach 2:
The patent changes the GOP (Group of Pictures) size parameter dynamically based on scene conditions. By adjusting the interval between I-frames according to motion detection results and scene complexity, the system optimizes bandwidth consumption while ensuring sufficient event detection coverage.
Data Source
AI summary
A system can obtain a labelled data set, including historic video data and labelled events. The system can divide the labelled data set into historic training/testing data sets. The system can determine, using the historic training data set, a plurality of different parameter configurations to be used by a video encoder to encode a video that includes a plurality of video frames. Each parameter configuration can include a group of pictures (“GOP”) size and a scenecut threshold. The system can calculate an accuracy of event detection (“ACC”) and a filtering rate (“FR”) for each parameter configuration. The system can calculate, for each parameter configuration of the plurality of different parameter configurations, a harmonic mean between the ACC and the FR. The system can then select a best parameter configuration of the plurality of different parameter configurations based upon the parameter configuration that has the highest harmonic mean.