An Intelligent Analytics for People Detection Using Deep Learning

People detection has become crucial in various applications, from security systems and surveillance to retail analytics and traffic management. With the advent of deep learning, particularly convolutional neural networks (CNNs), we’ve witnessed significant advancements in object detection accuracy and efficiency. This paper explores the power of intelligent analytics driven by deep learning for people detection, highlighting its benefits, challenges, and potential applications. The main aim is to build a people behaviour detection framework through body language, events, objects around people and their postures to determine the behaviour of people and environment genuinely based on given attributes like walking (still or moving), sitting (still or fidgeting), running (steady paise or high speed) and standing (still or fidgeting). These attributes contribute to detecting people’s behaviour from a given input of video sequence, both in real-time or pre-recorded from MATLAB using three different deep learning algorithms (CNN, You Only Look Once (YOLO) and Faster region CNN). The results obtained were compared to determine which model best suits people’s behaviour detection.