Human Activity Recognition a Comparison Between Residual Neural Network and Recurrent Neural Network

Recognizing human activity is important for interpersonal interactions and human-to-human communication. It is challenging to extract since it contains details about a person’s identity, personality, and psychological condition. One of the key research topics in the fields of computer vision and machine learning is the human capacity for activity recognition. This study has led to the need for a multimodal activity identification system in several applications, such as video surveillance systems, human-computer interaction, and robots for characterizing human behavior. In this study, Residual Neural Network (ResNet50) and Recurrent Neural Network (RNN) architectures for human activity recognition (HAR) are compared. Our assessment is based on a number of performance metrics, such as accuracy F1-score, recall, and computational effectiveness. Here we have the model accuracy of ResNet50 at 53% and RNN at 23% with epoch 10. Our objective is to evaluate how well the models can categorize various human activities. The outcomes demonstrate that ResNet50 outperformed RNN.