Human Activity Recognition (HAR) is a field of study that focuses on developing methods and techniques to automatically identify and classify human activities based on data collected from various sensors. HAR aims to enable machines like smartphones, wearable devices, or smart environments to understand and interpret human activities in real-time.
Traditionally, wearable sensor-based and camera-based methods were used. Wearable sensors are uncomfortable and inconvenient for users. Camera-based methods require intrusive installations, raising privacy concerns. Existing HAR technologies face challenges such as location dependency, sensitivity to noise, and a need for more flexibility in recognizing diverse activities in various applications, from smart homes to healthcare and the Internet of Things (IoT). The method used by UTeM provides a precise, adaptable, and location-independent solution.
University Teknikal Malaysia Melaka (UTeM) researchers have formulated an approach to Human Activity Recognition (HAR) to tackle traditional limitations. They have introduced a system that leverages Channel State Information (CSI) and advanced deep learning techniques.
The system uses Channel State Information (CSI) combined with Long Short-Term Memory (LSTM) networks. The system extracts vital indicators of wireless communication channel states, allowing for real-time classification and absolute location-independent sensing. LSTM networks facilitate sequential learning of activity features, simplifying the recognition process and accommodating variations in human activities across different people and environments.
The researchers emphasized that first data collection and preprocessing were conducted using Raspberry Pi 4 and specialized firmware to obtain raw Channel State Information (CSI) data, which was further improved using MATLAB for optimal quality and application.
Long Short-Term Memory (LSTM) networks were utilized to extract crucial features from the CSI data, thereby enabling the accurate recognition of complex human activities. They performed rigorous training on the LSTM model and classification process, which included an online phase for pattern recognition and an offline phase for enhanced performance.
The system introduces a signal segmentation method using the LSTM algorithm to determine the start and endpoint of human activities accurately.
Researchers tested the system and found that it achieved an impressive 97% accuracy rate in recognizing human activities. It showcased its capability to adapt to new environments, marking a significant advancement in HAR technology.
The researchers highlighted the remarkable adaptability of their system. It can easily blend into different settings without requiring extensive retraining or major changes. This flexibility makes it a practical solution for various fields, effectively addressing a wide range of real-world requirements. This approach represents a significant advancement in HAR technology and has the potential to significantly alter several industries, including smart homes, healthcare, and the Internet of Things.
Check out the Paper. All credit for this research goes to the researchers of this project. Also, don’t forget to join our 33k+ ML SubReddit, 41k+ Facebook Community, Discord Channel, and Email Newsletter, where we share the latest AI research news, cool AI projects, and more.
If you like our work, you will love our newsletter..
The post Deep Learning in Human Activity Recognition: This AI Research Introduces an Adaptive Approach with Raspberry Pi and LSTM for Enhanced, Location-Independent Accuracy appeared first on MarkTechPost.
#Applications #ArtificialIntelligence #ComputerVision #DeepLearning #EditorsPick #Staff #TechNews #Uncategorized [Source: AI Techpark]