Multi-Sensor Activity Classification With Location-Aware Feature Embedding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing human activity recognition systems fail to effectively integrate location information from multiple sensors, leading to suboptimal activity classification due to the lack of consideration for the unique properties and installation locations of different sensors.
Innovation Solution
The system represents sensors in a map-based feature embedding space, integrating location information with feature vectors from multiple sensors to create a composite feature vector that accounts for the specific properties and installation locations of each sensor, using deep convolutional neural networks for improved activity recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If location information from multiple sensors is not integrated, then the system complexity remains low, but the activity classification accuracy deteriorates
Solution Approach 1:
The patent combines location information from multiple sensors with feature vectors into a unified representation. Sensors are represented in a map-based feature embedding space where their locations and properties are integrated, creating a composite feature vector that merges spatial and sensory data for improved activity classification
Solution Approach 2:
The patent introduces a map-based feature embedding space as an intermediary representation. This embedding space serves as a mediator that transforms raw sensor locations and properties into a standardized format that can be effectively integrated with feature vectors from multiple sensors
2Measurement precision
If sensor properties and locations are considered, then the activity recognition precision improves, but the data processing complexity increases
Solution Approach 1:
The patent transforms sensor properties and locations into a standardized embedding space representation. By changing the parameter representation format to a unified embedding structure, the system can incorporate diverse sensor characteristics without proportionally increasing processing complexity
Solution Approach 2:
The patent represents sensors in a map-based feature embedding space that adds spatial dimensionality to the feature representation. This dimensional transformation allows location information to be naturally integrated with sensor properties in a structured manner
Data Source
AI summary
A method for classifying activity based on multi-sensor input includes receiving, from two or more sensors, sensor data indicating activity within a building, determining, for each of the two or more sensors and based on the received sensor data, (i) an extracted feature vector for activity within the building and (ii) location data, labelling each of the extracted feature vectors with the location data, generating, using the extracted feature vectors, an integrated feature vector, detecting a particular activity based on the integrated feature vector, and in response to detecting the particular activity, performing a monitoring action.