Video Object Detection Using Selective Classifier Application
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Methods for processing video data, such as identifying objects in a video, are computationally intensive, leading to high processing demands.
Innovation Solution
A computer-implemented method and system that processes video frames using a set of classifiers to generate feature data and classification data, allowing for efficient object detection by selecting a subset of classifiers based on previous frame analysis, reducing redundant processing and computational cost.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all classifiers are applied to every frame for object detection, then classification accuracy is improved, but computational cost and processing time increase significantly
Solution Approach 1:
The patent applies preliminary action by performing classification on the first frame using all classifiers to generate first classification data before processing the second frame. This preliminary classification results are then used to guide and optimize the processing of subsequent frames, allowing the system to prepare in advance for more efficient processing while maintaining accuracy.
Solution Approach 2:
The patent implements partial action by selectively applying classifiers to only certain portions of the second frame based on the first classification data. Instead of applying all classifiers to the entire second frame, the system processes only relevant regions with appropriate classifiers, reducing computational overhead while maintaining detection accuracy for objects of interest.
2Productivity
If a subset of classifiers is selected based on first frame data, then processing demands are reduced, but risk of missing objects increases
Solution Approach 1:
The patent applies feedback by using the first classification data obtained from the first frame to inform and guide the processing of the second frame. The classification results from the first frame provide feedback that helps select which classifiers to apply and which portions of the second frame to process, creating a closed-loop system that adapts based on previous results while maintaining reliability.
Solution Approach 2:
The system performs preliminary classification on the first frame to gather information about object presence and characteristics before processing the second frame. This preliminary action allows the system to make informed decisions about resource allocation for the second frame processing, ensuring that appropriate classifiers are applied to maintain detection reliability while optimizing efficiency.
Data Source
AI summary
A method of processing video data includes processing a first frame to generate first feature data, which is processed to generate first classification data. The first classification data is used to determine that a first class of object is present in the first frame. The first feature data is processed to determine a portion of the first frame in which an object of the first class is located. A portion of a second frame is determined in dependence on the determined portion of the first frame. A second set of classifiers is selected on the basis of the first classification data. The second frame is processed to generate second feature data. A portion of the second feature data is processed, using the second set of classifiers, to generate second classification data for determining whether the first class of object is present in the determined portion of the second frame.