TY - JOUR AU - Dr.A.Punitha,, G.Rajalakshmi, PY - 2021/10/07 Y2 - 2024/03/29 TI - Efficient Method For Identification Of Anomalies Based On A Deep Learning Classification Methodology JF - Design Engineering JA - DE VL - IS - SE - Articles DO - UR - http://thedesignengineering.com/index.php/DE/article/view/5150 SP - 2331-2350 AB - This research intends to enhance the identification problem in visual processing using the deepanomaly identification system. The Backdrop Estimation (BE)module, an Objective Separation (OS) module, an Extraction of Features (FE) module, and an Activities Recognizing (AR) module are provided in the system to complete the identification of anomaly. An anomaly detection system based on the deep learning classification (ADS-DLC) model is proposed in this research. It initially provided aBackdrop Estimation (BE) module that produced an exact background for generating two frameworks to calculate the background assessment.The OS architecture is employed to collect the target from videos after generating a quality backdrop. The target tracking method is then utilized to track the item using the coinciding detection program. The FE component is removed from the monitored objects for relevant form, frequency, and distribution for unusual action recognition. The deep learning classification categorizes the suggested AR module as an unusual or usual occurrence for the last phase.Investigations are carried out with the benchmarks of aberrant operations, and the approach's benefits are validated with the most advanced methods. It can observe that the suggested activity identification system was superior to the current systems by attaining better simulation outcomes. It also indicates that the proposed technique achieves 91 percent of the structure level of performance. ER -