In the modern-day, automatic surveillance is an active research issue. It has taken the command in many fields and many remain untouched. Our main purpose is to detect anomalies at different places using CCTV. The importance of detecting suspicious human activity via video surveillance is to stop theft cases. For detecting abnormal actions we will be using Convolutional Neural Network(CNN), Neural Learning, System Learning, Long short-term memory (LSTM) and Deep Structured Learning for detection of abnormal actions of students on campus. The use of artificial intelligence enables computers to think like people. Making predictions based on future data and learning from training data are crucial aspects of machine learning. Further, Deep learning is employed since there are now GPU (Graphics Processing Unit) processors and large datasets accessible. In previous research LSTM was used which was not so effective, therefore Long-term recurrent convolutional network (LRCN) training model will be used here to increase the accuracy of the anomaly detection. The proposed model will be able to detect human normal and suspicious activities based on the training model