Advancing Activity Recognition With Multimodal Fusion and Transformer Techniques
Aidarova S. Nurmakhan T. Myrzakhan R. Fazli S. Yazici A.
2025Institute of Electrical and Electronics Engineers Inc.
IEEE Sensors Journal
2025#25Issue 1119632 - 19649 pp.
In the field of human activity recognition (HAR), the precise identification of human activities from time-series sensor data is a complex yet vital task, given its extensive applications across various industries. This study introduces an advanced HAR technique that markedly improves the activity recognition by combining multimodal sensor fusion with a Transformer-based attention mechanism. Our methodology begins with rigorous preprocessing of the raw data from multiple sensors, focusing on cleaning and normalizing the data to create ideal conditions for subsequent analysis. We then apply an innovative sensor fusion strategy alongside a Transformer-based attention mechanism to accurately and comprehensively detect human activities. The effectiveness of our method was rigorously evaluated on two widely recognized benchmark datasets in HAR research, Extrasensory, and UCI-HAR, both known for their complexity and broad usage. The evaluation concentrates on the model’s ability to precisely classify primary human activities, particularly in near-real-time situations. The findings demonstrate that our model excels in accuracy and adeptly identifies primary human activities, marking a notable progression in HAR technology. This enhancement highlights the strong potential of our proposed approach, combining sensor fusion and attention mechanisms to advance activity recognition in real-life settings.
Attention mechanism , human activity recognition (HAR) , sensor data fusion , smartphone data , Transformer
Text of the article Перейти на текст статьи
Nazarbayev University, School of Engineering and Digital Sciences, Department of Computer Science, Astana, 010000, Kazakhstan
Nazarbayev University
10 лет помогаем публиковать статьи Международный издатель
Книга Публикация научной статьи Волощук 2026 Book Publication of a scientific article 2026