Deep Learning-Based Weapon Detection in Video Streams with Improved LSTM (I-LSTM) Models

This research study introduces a sophisticated method for detecting weapons in video streams by utilizing deep learning techniques with enhanced Long Short-Term Memory (LSTM) models. The suggested approach employs a fusion of pre-trained Convolutional Neural Networks (CNNs) to extract features and Improved LSTM models to capture temporal relationships in video sequences. The experimental results exhibit a higher level of accuracy when compared to the existing models in weapon detection scenarios. The study entails a thorough assessment utilizing a varied dataset of video frames that have been annotated to indicate whether weapons are present or not. The Enhanced LSTM model demonstrates superior performance, attaining greater levels of accuracy, precision, recall, and F1 score metrics in comparison to traditional LSTM models and other current methodologies. Integrating Improved LSTM not only enhances the accuracy of detection but also enhances the resilience to the time-based changes that naturally occur in real-world video streams. This research enhances the progress of deep learning methods for security purposes by providing a dependable and effective solution for identifying weapons in dynamic video settings. The experimental results demonstrate the efficacy of the suggested model, suggesting its potential for practical use in surveillance and security systems.