Human activity detection and classification by AI approaches
Anish Thakur, Geet Kiran Kaur, Ranjit Singh · Computational Methods in Science and Technology · 2024
The real area where it is well researched is Human Activity Recognition in the field of Computer Vision. Computer vision is one of the well-explored areas of interest in the proper processing and analysis of visual information with respect to specified applications, e.g., security, health care, human-computer interaction. This will be a huge step forward in effective identification and categorization of human behaviors based on sensor data due to the development of machine/deep learning techniques. Within the present paper outline, there is an attempt to provide a comprehensive review of recent work and advances in the area of human activity detection based on machine/deep learning approaches. This task will review the major challenges in human activity detection, covering dealing with several activities, sensor data unpredictability, and reliable methodologies for classification. We have thus analyzed these models through machine learning-based techniques: traditional, such as Support Vector Machines (SVM) and Random Forests, and innovative deep learning architectures like Recurrent Neural Networks (RNNs) and Convolutional Neural Networks (CNNs). It also covers the problems of small data, challenges for model interpretability, and difficulties to use the models for its purposes in real time. Next, it also gives a glance at potential use cases under health, sport data analysis, and intelligent environments.