Anomaly Detection in Video Surveillance Using Convolution Neural Network
Kallepalli Rohit Kumar, Nisarg Gandhewar · 2022
Day by day world is moving towards the digital platform, and in the digital world, real-time video surveillance systems are all in high demand because of the expanding needs of cities and industries. So far, many conventional methods have been implemented in this era, but nowadays, machine learning and artificial intelligence-based anomaly identification system address a portion of the difficulties. AI and machine learning-based methodology extract the image’s dynamic features (aberrant or abnormal behavior across time). The constant movement of people and the fluctuating conditions of outdoor cameras, along with dimensionality and features in video frames, make anomaly detection very challenging. This article uses the neural network architecture to extract the features from the input video and identify the anomalies in video frames. Finally, we used the CUHK Avenue dataset to authenticate the proposed work. According to the results, the outcomes of the proposed work are over performed compared to other state-of-the-art methods.