Mobile Terminal Pocket State Detection Using Neural Network Classifier
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining whether a mobile terminal is in a pocket suffer from low accuracy, particularly in scenarios where the device is in a moving vehicle or when the distance sensor is covered, leading to incorrect determinations.
Innovation Solution
A state detection method that involves acquiring a state sample set from multiple mobile terminals, training a classifier using this data, and sending it to the mobile terminal to determine if it is in a preset holding space based on various factors such as proximity distance, acceleration, system time, ambient light, and program running status.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If individual sensor features (distance sensor and acceleration sensor) are used to determine pocket state, then the device complexity is low, but the measurement precision is low leading to incorrect determinations in specific scenes
Solution Approach 1:
The patent combines multiple sensor features (distance sensor, acceleration sensor, light sensor, proximity sensor) and integrates them through a neural network model to form a unified detection system. This merging of multiple data sources enables accurate pocket state determination in complex scenarios where individual sensors fail, directly resolving the contradiction between low measurement precision and low device complexity.
Solution Approach 2:
The neural network model serves as an intermediary that processes and integrates data from multiple sensors. Instead of directly combining sensor outputs, the neural network learns optimal feature combinations and relationships, acting as a smart mediator that transforms individual sensor readings into accurate pocket state predictions while handling the complexity internally.
2Measurement precision
If multiple sensor parameters and neural network classification are implemented, then the measurement precision improves, but the use of energy increases due to complex processing
Solution Approach 1:
The neural network model is trained offline in advance using labeled data from multiple sensors. This preliminary action transfers the computationally intensive learning process to a separate training phase, allowing the mobile device to only perform lightweight inference during actual operation. The model is pre-configured with learned parameters that enable accurate classification without requiring heavy real-time computation, thus reducing energy consumption during deployment.
3Reliability
If traditional methods are used to determine pocket state, then the ease of operation is high, but the reliability is low due to false positives in moving vehicles
Solution Approach 1:
The system implements feedback mechanisms where the neural network continuously receives input from multiple sensors and adjusts its predictions based on combined evidence. The model learns from patterns in the data, including scenarios like moving vehicles, and provides feedback-driven corrections to avoid false positives. This feedback loop enables reliable determination while maintaining ease of operation through automated decision-making.
Data Source
Figure 1~2A
Figure 2B
Figure 3A
AI summary
The present disclosure provides a method and an apparatus for detecting a state and a storage medium, belonging to a field of mobile technology. The method may include: acquiring (201) a state sample set of a plurality of sample mobile terminals, determining (202, 304) a classification algorithm; acquiring (203, 305) a classifier by training the state sample set determined as training data according to the classification algorithm; and sending (204, 306) the classifier to a mobile terminal to be detected, in which the mobile terminal to be detected is configured to determine whether the mobile terminal to be detected is located in the preset holding space based on the classifier and according to a state sample of the mobile terminal to be detected.