In recent years, there has been a growing interest in video anomaly detection systems. While current approaches rely on deep learning techniques for this task, there are several inherent challenges associated with this approach. Deep learning methods, in general, are susceptible to issues like noise, concept drift, lack of explainability, and the need for large training datasets. Moreover, anomaly detection is a complex problem, characterized by challenges such as dealing with unknown anomalies, handling data heterogeneity, and addressing class imbalance. To tackle these challenges, deep learning-based anomaly detection methods often resort to unsupervised generative models like generative adversarial networks (GANs) and autoencoders. However, these models are not immune to the broader issues associated with deep learning and can be challenging to train effectively. In the proposed system, the potential of the hierarchical temporal memory (HTM) algorithm for video anomaly detection is investigated. HTM offers certain advantages. To address the unique requirements of video anomaly detection, a novel variant of HTM called “Grid HTM,” tailored specifically for this purpose is proposed. The anomaly score, which represents the likelihood or degree of an anomaly within a dataset or system, is displayed in the pickle file. The file displays an anomaly score of each grid (between 0 and 1). The higher the score, the higher the anomaly. Furthermore, for more readability, it has been converted into JSON format and for a visual output, into a histogram.