A Review on Time Series Learning-Based Human Action Recognition

In recent years, the interest in human activity recognition (HAR) has increased due to its importance in various fields including medicine When time series features are extracted from sensor data, they are important for humans equitable distribution of activities. However, traditional methods are based on heuristic handcrafted features and shallow learning architectures, which do not capture the specific features necessary for classification the complex human activity poses challenges in extracting effective features from sensor data. Deep learning has emerged as a promising approach to overcome these challenges, overcoming the obstacles of time series extraction and classification. However, challenges such as computational complexity and data access remain. The aim of this study is to describe HAR based on time series.