Abstract & Overview
Video anomaly detection is one of the trickiest issues in intelligent video surveillance because of the complexity of real data and the hazy definition of anomalies. Since abnormal occurrences typically seem different from normal events and move differently. The global optical flow was determined with the maximum accuracy and speed using the Farneback approach for calculating the magnitudes. Two approaches have been used in this study to detect strangeness in the video. These approaches are Deep Learning (DL) and manuality. The first method uses the activity map's development of entropy to detect the oddity in the video using a particular threshold. The second method uses a Convolutional Recurrent Auto Encoder (CRAE). CRAE is a network that combines a convolutional autoencoder and an attention-based Convolutional Long-Short-Term Memory (ConvLSTM) network. The irregularity regarding the temporal pattern and the spatial irregularity, respectively, might be captured by the convolutional autoencoder and ConvLSTM network. The current output properties of each CovnLSTM layer were extracted from their hidden states using the attention method. Comparing the error with an experimentally established threshold, anomalies were specified to exist and a convolutional decoder was used to recreate the input video clip and the testing video clip. The best detection of whether in-frame variation was abnormal or normal, a trial-and-error threshold was 0.04 for handcrafted features through the University of Minnesota (UMN) dataset and 0.00035 for DL features through the avanue dataset.