Action Recognition Using Key Point Visibility Attributes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional technologies are inadequate in recognizing complex actions of target objects due to inaccurate determination of visibility attributes of key points, leading to incomplete action recognition.
Innovation Solution
A method and device that acquire an original image, recognize the target object, detect key points, determine visibility attributes, and recognize actions based on a combined value of these attributes, enabling accurate identification of actions by using a key point detecting model and a visibility determining model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods determine action based on simple state changes of target object, then simple actions can be recognized, but complex actions cannot be recognized accurately
Solution Approach 1:
The patent segments the target object into multiple key points (e.g., hand keypoints) and evaluates visibility of each key point individually. This segmentation allows the system to capture complex spatial relationships and occlusion patterns that simple state changes cannot detect, thereby enabling accurate recognition of complex actions while maintaining coverage of simple actions.
2Measurement precision
If visibility attributes of key points are not determined, then processing is simpler, but action recognition is incomplete
Solution Approach 1:
The patent performs preliminary determination of visibility attributes for each key point before action recognition. By pre-calculating whether each key point is visible or occluded and storing these attributes, the system prepares essential information in advance that enables accurate complex action recognition without adding excessive complexity during the main recognition process.
Data Source
AI summary
An action recognition method and device for a target object, and an electronic apparatus is disclosed. The action recognition method for a target object comprises: obtaining an original image from an image source, the original image comprising a target object, recognizing the target object from the original image; detecting a plurality of key points of the target object; determining, on the basis of the detected key points, visibility attributes of the plurality of key points, wherein the visibility attributes are used to indicate whether or not the key points are occluded; and recognizing an action of the target object according to a combined values of the visibility attributes of the plurality of key points. The method uses visibility of key points to determine actions of a target object, thereby resolving the technical issue in the prior art in which complex actions cannot be accurately recognized.