Anomaly Detection Approach for Human Detection in Crowd Based Locations

Currently, it is challenging to find a solution for human identification in a busy area. To deal with security issues like theft, fire, or other strange incidents, private organizations also install security cameras on their property. Deep learning has now proven to be effective in a wide range of industries, from audio and video to NLP and image recognition. For Deep Learning techniques like Convolutional Neural Networks (CNNs), Graph Neural Networks (GNNs), etc., abnormal event detection is a typical application case. Modern 3D convolutional neural network (CNN)-based models for anomaly detection will be used in this study. The Avenue collection's footage of both common and atypical crowd behavior is used in experiments. By comparing the percentage of correctly detected frames to the actual data, the model's efficacy is evaluated. In this paper, we present a reliable method for spotting unusual crowd behavior. The input from numerous sensors used by various models is used to calculate accuracy. Finding outliers and giving comparable estimates is the main objective. The proposed model has the maximum accuracy (92.6%), according to the findings of the trials that were carried out.