LITERATURE REVIEW ON CROWD COUNTING AND CROWD DENSITY MAPPING METHODOLOGIES

Abstract

Purpose of Review: Artificial Intelligence and Machine Learning technologies have enabled the analysis of the crowd which helps monitor and manage the crowd effectively. With further advancements in the field of AI and ML, the quality and the accuracy of the analysis have improved considerably. This review attempts to provide an overview of the currently available technologies enabling crowd analysis operations like crowd counting, crowd density mapping and weigh the benefits and drawbacks of the available methods also take into consideration the situation they are suitable for. Recent Finding: We have reviewed 10 articles related to crowd counting and crowd density mapping that used Multi-Column CNN, Deep Convolutional Neural Networks, Real-Time Deep network, Switching CNN, andEncoder-Decoder Based CNN with Multi-ScaleAware Modules to estimate the crowd density and count for the given crowd scene image. In summary, these studies showed a high level of accuracy and provided excellent examples of the potential of AI in crowd counting. The methodologies and algorithms adopted to handle the problem of crowd counting and density mapping varied considerably from paper to paper. Few Built upon the findings of older methodologies by adding enhancement layers and others approached the problem with a new perspective. The methodologies we studied, had their advantages and disadvantages and were suitable for certain situations. Summary: With advancements in Computer Vision and Artificial neural network techniques like Convolution Neural Networks, crowd counting, and density mapping which traditionally were considered a difficult task and take up time to train the models has become easier to implement and time for training the models are also reduced considerably reduced. In this paper, we have surveyed journals and articles from a variety of sources via google scholar published from 2015 to 2021 related to crowd counting and crowd density mapping. We have reviewed the methodologies introduced in this paper along with their advantages and disadvantages.