Advancing Event Recognition in Videos Using Hybrid Deep Learning Model
H Jyothi, M Komala, S Mallikarjunaswamy · 2024
Event recognition in videos plays a critical role in various applications such as surveillance, security, and human-computer interaction. Conventional methods like Hidden Markov Models (HMM) and Support Vector Machines (SVM) have been extensively used, but they often struggle with handling complex data patterns and large-scale video datasets, leading to limited accuracy and inefficiency in real-time scenarios. These approaches fail to capture intricate temporal dependencies and suffer from high computational costs, making them less effective for dynamic event detection. To address these limitations, we propose a Hybrid Deep Learning Event Recognition Model (HDL-ERM), which combines the strengths of Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks. CNN is used for feature extraction, while LSTM effectively captures temporal sequences in video data. The HDL-ERM model significantly improves event recognition accuracy by 0.25%, while reducing computational load by 0.18% compared to conventional methods. By leveraging deep learning techniques, our proposed model enhances performance across various parameters such as processing speed, detection accuracy, and adaptability to different video environments.