Machine learning has been increasingly important all over the world due to its powerful abilities in numerous areas. It has shown great potential in the scientific community, such as chemistry and medicine, and other fields, like economics. So far, human activity recognition (HAR) is also becoming essential in daily lives, such as robot interactions and health monitoring devices. By combining machine learning, especially the technique of convolutional neural networks (CNNs), and HAR, the efficiency and accuracy of HAR will significantly improve, and create more opportunities for future research and development. There are already several existing areas for deep learning-based HAR, such as self-driving cars and fitness. The purpose of this paper is to organize and review the current progress on deep learning-based HAR techniques, and major applications of it in everyday lives. There are still some challenges in the current techniques, but as more studies and research are conducted in the near future, deep learning-based HAR techniques will be greatly improved and become a critical part in the real world.
Yucheng Sheng (Mon,) studied this question.