A Study on Human Activity Recognition using CNN and LSTM

This paper aims to develop a robust Human Activity Recognition (HAR) system by integrating Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks, leveraging the UCF 50 dataset. The UCF 50 dataset, which includes 50 di- verse human action categories from real-world videos on YouTube, presents a significant challenge for accurate activity recognition. The hybrid model combines CNNs for spatial feature extraction and LSTMs for capturing temporal dependencies, enabling the recognition and classification of complex human activities over video sequences. The study seeks to improve recognition accuracy, offering valuable applications in areas such as surveillance, human-computer interaction, and sports analytics.