Violence Detection Alert System Using Deep Learning with SMTP

Violence detection alert systems ensure the safety and security of the public. By utilizing deep learning techniques, we have developed a system for detecting violent behaviour in areas that require a security system, such as crowded environments. The system analyses the data using convolutional neural networks and recurrent neural networks. To begin with, CNN is used to extract spatial features from the video, which are then classified as violent or non-violent based on the actions. Next, RNN is introduced to analyse and identify the instances of violence in video frames. Moreover, a Bi-LSTM is integrated into this system to enhance the model and improve accuracy. Additionally, the mechanism focuses on key areas within each frame to enhance the model’s accuracy and capabilities. The dataset comprises both violent and non-violent videos, enabling high performance across a wide range of situations. To achieve high performance, our experiments demonstrate that the model performs well, achieves good accuracy, and efficiently detects violence in any setting. To make the system more practical, we have utilised SMTP in this system, which will send an alert message via email when violence is detected, allowing for timely intervention.