Face and Object Detection Algorithms for People Counting Applications

Thisresearch work delves into the significance of people counting systems, emphasizing their role in furnishing valuable data for operational improvement, security enhancement, and resource optimization in various businesses and organizations. The focus is on a thorough examination of face detection and object detection methodologies rooted in computer vision and deep learning. Specifically, the study scrutinizes Multi-task Cascaded Convolutional Networks (MTCNN) and YuNet for face detection, along with the Histogram of Oriented Gradients (HOG) feature descriptor coupled with Support Vector Machine (SVM) and the real-time capabilities of You Only Look Once version 8 (YOLOv8) for object detection. Through empirical evaluation across diverse conditions and comparative analysis, this research aims to elucidate the strengths and limitations of these algorithms in the context of people counting tasks. The findings provide valuable insights to guide the selection of suitable approaches for specific use cases, contributing to the ongoing progress in the field of computer vision.