Temporal Action Detection for Arbitrary Duration Resources
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current temporal action detection methods are limited in their ability to apply to resources of arbitrary duration and lack strong frame mobility, making them less effective for long-duration tasks and various types of resources.
Innovation Solution
A method that performs feature extraction on target frame data to obtain feature data, followed by temporal action proposal to generate candidate action proposals, and then classifies these proposals to determine action detection results, including category and period, allowing for action detection in resources of any duration and type.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If current temporal action detection methods are used, then action detection can be performed on standard resources, but the method is limited in its ability to apply to resources of arbitrary duration and lacks strong frame mobility
Solution Approach 1:
The method segments the action detection process into three independent modules: feature extraction, temporal action proposal, and classification. This segmentation allows each module to process data of arbitrary duration independently, improving adaptability while maintaining detection reliability through modular processing
Solution Approach 2:
The method dynamically adjusts the temporal action proposal mechanism to handle resources of varying durations. By using a proposal network that can generate action intervals adaptively based on input characteristics, the system achieves strong frame mobility and reliability across different resource types and lengths
2Measurement precision
If feature extraction is performed on target frame data, then action detection accuracy is improved, but processing time and computational complexity increase
Solution Approach 1:
The method performs preliminary feature extraction on target frame data before action proposal and classification. By pre-extracting and storing feature data in a cache structure, the system prepares computational resources in advance, reducing real-time processing time while maintaining high detection accuracy
Solution Approach 2:
The method maintains continuous feature extraction and caching operations throughout the video processing pipeline. By keeping the feature extraction process continuous and reusable, the system avoids redundant computations and maintains accurate action detection while optimizing processing efficiency
Data Source
AI summary
A method of detecting an action, an electronic device, and a storage medium. A method can include: performing a temporal action proposal on at least one target feature data obtained by a feature extraction on a plurality of target frame data of a target resource, so as to obtain at least one first candidate action proposal information; classifying target feature data corresponding to at least one first candidate action proposal interval included in the first candidate action proposal information, so as to obtain at least one classification confidence level corresponding to the at least one first candidate action proposal interval; and determining an action detection result for at least one action segment contained in the target resource according to the at least one classification confidence level corresponding to the at least one first candidate action proposal interval, wherein the action detection result includes an action category and an action period.