Comparative Study of Deep Learning Based Methods for Human Activity Recognition

Recognizing human activity is essential to improving interpersonal relationships and interactions between people because it provides important insights about a person's identity, personality, and psychological condition.The difficulties in the field are shown by how difficult it is to appropriately extract this information.Research in artificial intelligence and machine learning is primarily focused on the ability of humans to perceive actions, which is propelling developments in a wide range of applications.Robust activity recognition systems are required for a variety of applications, including robots, human-computer interaction, and video surveillance.Recognition of human activity has become an important area of research in image and video analysis.Numerous research has examined this subject over time, emphasizing both its importance and the continuous search for better approaches.In this regard, we suggest a unique method to improve the accuracy of human action and activity recognition using OpenCV, Convolutional Neural Networks, and Graph Neural Networks based on deep learning.Our approach uses CNN and GCN algorithms, which are excellent at finding features and patterns in pictures and video sequences, to train a large dataset.This is enhanced by OpenCV, a potent real-time computer vision technology, which makes the identification system's implementation easier.Our method seeks to achieve high precision in identifying and categorizing different human actions by combining CNN, GCN and OpenCV, advancing fields like security, UI design, and autonomous systems.