Date of Publication :2nd September 2024

Abstract:This paper presents an innovative machine learning approach designed to accurately determine the density of a crowd depicted in an image. Our primary objective is to classify images into different density categories, ranging from low to high, using a combination of image features and advanced machine learning techniques. We then apply specific methods to each density category to achieve optimal crowd size estimation. Furthermore, we propose a framework for classifying crowd density into three levels: low, medium, and high. This framework utilizes a combined CNN and RNN(LSTM) approach, taking into account factors such as security and individual comfort. It holds promise for security agencies responsible for crowd management, aiding in the prevention of crowd-related incidents. Additionally, it offers valuable insights for organizations tasked with providing accurate crowd attendance data for public events.

Reference :