Chronological ant lion optimizer-based deep convolutional neural network for panic behavior detection in crowded scenes

Nowadays, crowd scene becomes the most active-oriented research and trendy topic in computer vision applications. However, panic behavior is the key suggestion of occurrence of the abnormal behavior in the human crowd such that detecting the panic behavior helps to prevent the disastrous situations. Various existing methods are adopted for detecting panic behavior in crowded scenes, but it results in performance degradation due to the varying density of objects in the crowded scenes. Hence, an effective method is developed using the proposed Chronological-Ant Lion Optimizer-based Deep Convolutional Neural Network (Chronological ALO-based Deep CNN) to detect the panic behaviors in crowd scenes. Nevertheless, the adopted Chronological ALO is modeled by incorporating the Chronological event with ALO and is mainly used for training the Deep CNN classifier in order to reveal better detection results. Based on the extracted feature, variations in the crowd dynamics are examined in the non-panic situation for characterizing the usual behavior of pedestrians. Finally, panic behavior is presumed as the deviation from disparities observed while non-panic situations. However, the proposed Chronological ALO-based Deep CNN obtained better performance with the metrics, like accuracy, sensitivity, and specificity with the values of 95.833%, 96.296%, and 96.329%, respectively.