Human behavior recognition has become an important research area in computer vision due to its applications in surveillance, healthcare monitoring, and human–computer interaction. Recognizing human actions from video data is challenging because it requires understanding both spatial features from individual frames and temporal relationships between them. Traditional methods based on handcrafted features often fail to capture these complex patterns effectively. In this study, a hybrid deep learning model combining Convolutional Neural Networks (CNN) and Long ShortTerm Memory (LSTM) networks is proposed for human activity recognition. The CNN extracts spatial features from video frames, while the LSTM captures temporal dependencies across frame sequences to improve action classification. The model utilizes MobileNetV2 as the CNN backbone for efficient feature extraction. The proposed system supports both offline video classification and real-time activity recognition using a webcam. Experimental results show that the CNN–LSTM architecture provides accurate and efficient recognition of human behaviors, making it suitable for practical applications such as intelligent surveillance and smart monitoring systems.
Talha et al. (2026) studied this question.