Video Playback Start Control Using Dynamic Buffer Frame Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The issue of poor video playback start experiences due to fixed threshold values for video buffer settings, leading to inconsistent performance across different network conditions, causing prolonged startup times and frequent freezing.
Innovation Solution
A method utilizing a pre-trained gradient boosting decision tree (GBDT) regression model to dynamically predict an optimized number of video buffer frames based on acquired feature data, such as network speed and video characteristics, to determine when to start playback.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a fixed threshold value for video buffer frames is used, then the video playback start can be controlled with a simple standard, but the video playback start speed becomes slow and freezing occurs under varying network conditions
Solution Approach 1:
The patent transforms the static fixed threshold value into a dynamic prediction mechanism. The GBDT regression model dynamically predicts the optimal video buffer frame threshold based on real-time feature data including network conditions, device performance, and video characteristics. This dynamic adjustment resolves the contradiction by allowing the system to adapt to varying network conditions while maintaining simple operational control through automated prediction.
Solution Approach 2:
The patent changes the parameter from a fixed threshold value to a predicted value that varies based on multiple input features. The GBDT model processes features such as network speed, device CPU performance, and video bitrate to output an optimized buffer frame threshold. This parameter transformation enables the system to achieve faster startup times under good network conditions while preventing freezing under poor conditions, resolving the time loss issue.
2Device complexity
If a fixed threshold value for video buffer frames is used, then the system structure remains simple, but the video playback performance becomes inconsistent across different network conditions
Solution Approach 1:
The patent implements a feedback mechanism where the GBDT model continuously predicts optimal buffer thresholds based on current network conditions and device states. The system collects feature data during playback, feeds it to the prediction model, and adjusts the buffer threshold accordingly. This feedback loop maintains system reliability across different network conditions while keeping the control structure relatively simple through automated decision-making.
Solution Approach 2:
The GBDT regression model acts as an intermediary between the complex network conditions and the simple buffer control decision. Instead of directly managing complex network variations, the model translates multiple input features into a single predicted threshold value. This intermediary approach maintains simple system structure while achieving reliable playback performance through data-driven predictions.
3Reliability
If the video buffer threshold is increased to prevent freezing, then playback smoothness improves, but the video startup time increases due to longer buffering
Solution Approach 1:
The patent applies preliminary action by pre-training the GBDT regression model with historical data to learn optimal buffer thresholds for various network conditions. During actual playback, the pre-trained model quickly predicts the appropriate threshold without requiring extensive real-time computation. This preliminary preparation enables the system to make fast, accurate decisions that prevent freezing while minimizing buffering time, resolving the contradiction between smoothness and startup time.
Data Source
AI summary
A method and apparatus for optimizing a video playback start, a device and a storage medium are provided. An implementation of the method may include: acquiring feature data, the acquired feature data affecting the video playback start speed when a video starts to play; inputting the acquired feature data into a pre-trained gradient boosting decision tree (GBDT) regression model to output a predicted value for a video buffer frame, and recording the predicted value as a first predicted value; and ascertaining whether to start playing the video based on the first predicted value.