Automated Incongruity Recognition Model Using Optimized Ensemble Pattern Extraction Condensation and Classification

Abstract

Crowd video analysis has attained more interest in the computer vision newlineapplications that includes the monitoring operation of human activities. Numerous newlinecrowded video frames provide diverse complexities in evaluating the videos that cannot newlinebe managed with the conventional approaches. The processing of surveillance video is newlinemainly considered for the applications of real-time to identify the unusual actions or newlineanomaly identification to offer management and security. But, the surveillance videos newlinehave a high amount of information that consumes more manpower and time to execute. newlineHence, manual execution is very troublesome. Hence, the computation approaches are newlinesignificant for supporting the security of humans in managing the crowded scenes. newlineMoreover, the video condensation is also employed in the anomaly identification process. newlineIt can reduce the length of the given videos and enrich the performance of the video. It newlinealso minimizes the execution time and decreases the computational cost. However, the newlineexisting video condensation processes are diminishing the quality of the video and also newlineinclude background issues such as noise and lightning. The anomalies in the videos are newlinethe abnormal activities or events that specify the undesired activity. Anomaly newlineidentification in the videos can result in the evaluation of the spatial and temporal newlineanomalies in the information. This anomaly leads to high false alarm rates and degrades newlinethe outcome of the model. Various computation mechanisms are implemented to identify newlinethe unusual activities in the crowded videos. The techniques detect the overall features newlineand significant object features in every frame. However, these approaches take more time newlineand label the information newline