Research on mobile machine learning platforms for human gesture recognition in human-machine interaction systems
Open Access
- 28 March 2025
- journal article
- Published by Private Company Technology Center in Technology audit and production reserves
- Vol. 2 (2(82)) , 6-14
Abstract
The subject of this research is mobile machine learning platforms for human gesture recognition within human-machine interaction systems, specifically for managing smart home components. One of the key challenges in gesture recognition is ensuring high accuracy, efficiency, and robustness of algorithms under real-world operating conditions. The problem lies in selecting optimal machine learning platforms capable of balancing local and cloud computing, processing speed, and adaptability to changing environmental conditions. The study presents a comparative analysis of the ML platforms Create ML (Apple) and Google Cloud AI Platform, which are used for gesture detection and recognition in smart home control systems. The obtained results demonstrate that Create ML achieves an accuracy of 95.81 %, while Google Cloud AI Platform reaches 89.43%, justifying their selection for further research. Additionally, experimental testing of sensor placement topology revealed that diagonal camera positioning increases accuracy by 0.62 % compared to parallel placement. The increased efficiency of Create ML is due to its ability to process data locally, reducing latency and dependence on an internet connection. In contrast, Google Cloud AI Platform relies on cloud resources, enabling the processing of large volumes of data but making it dependent on data transmission speed. The proposed gesture control algorithms can be used to enhance the accessibility of technology for people with disabilities, particularly in rehabilitation centers. Additionally, the research findings can be applied to contactless interfaces in medical facilities and public spaces, reducing the need for physical interaction with surfaces and improving hygiene levels. The use of mobile ML platforms in such scenarios allows for the optimization of computational resources and ensures the effective integration of gesture control into modern human-machine systems.Keywords
This publication has 24 references indexed in Scilit:
- Big AI: Cloud infrastructure dependence and the industrialisation of artificial intelligenceBig Data & Society, 2024
- A Systematic Review of Hand Gesture Recognition: An Update From 2018 to 2024IEEE Access, 2024
- An Empirical Study of Challenges in Converting Deep Learning Models2022
- Face Mask Detection in Realtime Environment using Machine Learning based Google Cloud2022
- Enhanced Intelligent Smart Home Control and Security System Based on Deep Learning ModelWireless Communications and Mobile Computing, 2022
- Human-Machine Interaction Sensing Technology Based on Hand Gesture Recognition: A ReviewIEEE Transactions on Human-Machine Systems, 2021
- Smart home security: challenges, issues and solutions at different IoT layersThe Journal of Supercomputing, 2021
- Emerging Wearable Interfaces and Algorithms for Hand Gesture Recognition: A SurveyIEEE Reviews in Biomedical Engineering, 2021
- Fast Adapting Without Forgetting for Face RecognitionIEEE Transactions on Circuits and Systems for Video Technology, 2020
- OUHANDS database for hand detection and pose recognition2016