Pose Estimation Neural Network for Limited Computing Capacity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current human pose estimation algorithms are not suitable for devices with limited computing capacity, leading to low accuracy and inability to implement applications that rely on these algorithms effectively.
Innovation Solution
A method and apparatus for pose estimation that includes digitally capturing an image, estimating object poses, obtaining skeleton information, and processing it for detecting occlusion, pose detection, and content adjustment using a lightweight neural network structure, specifically a feature extraction neural network and a backend prediction neural network, optimized for devices with limited computational capacity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If existing human pose estimation algorithms are used, then pose estimation capability is provided, but the algorithms cannot be implemented on devices with limited computing capacity such as terminal devices
Solution Approach 1:
The patent segments the pose estimation algorithm into distinct functional modules: feature extraction module, pose parameter calculation module, and application module. This segmentation allows the algorithm to be implemented in a modular fashion on terminal devices with limited computing capacity, reducing the burden on any single component while maintaining overall functionality.
2Measurement precision
If existing human pose estimation algorithms are implemented on terminal devices, then pose estimation is possible, but the accuracy of pose estimation is insufficient to satisfy high level requirements in applications
Solution Approach 1:
The patent changes key parameters of the pose estimation algorithm to optimize for both accuracy and speed on terminal devices. This includes adjusting feature extraction parameters, pose calculation thresholds, and processing iterations to achieve high accuracy while maintaining processing speed suitable for real-time applications on devices with limited computing capacity.
3Measurement precision
If high-accuracy pose estimation is achieved, then application requirements are satisfied, but the computational load increases making it difficult to implement on terminal devices
Solution Approach 1:
The patent extracts and implements only the essential feature extraction and pose calculation functions needed for high-accuracy pose estimation on terminal devices. By taking out and implementing only the critical components rather than using complete complex algorithms, the system achieves high accuracy while significantly reducing computational energy consumption to levels suitable for terminal devices.
Data Source
AI summary
Provided is a method for pose estimation in a device, the method comprising capturing an image; estimating poses of an object included in the captured image; obtaining skeleton information of the object based on the estimating of the poses of the object; and processing the skeleton information of the object for at least one of detecting blocking of the object, detecting the poses of the object and adjusting content based on detected virtual object distinct from human body poses.