Comparative Study of Mask R-CNN and PointRend for Surveillance Video Human Segmentation: Accuracy, History, Evaluation Datasets, Challenge Classification, and Future Prospects

Computer vision's video segmentation is essential to many applications. This study examines the Mask R-CNN and PointRend algorithms’ performance and constraints in segmenting humans in surveillance video. History of segmentation methodologies and a detailed analysis of the two algorithms are presented in the study. Comparisons are made of their functionality, performance, salient features, and pros and cons. The research also reviews Mask R-CNN and PointRend datasets for human segmentation, benefiting other researchers. The study also examines Mask R-CNN and PointRend's origins, functions, and significant con-tributions to video human segmentation. It systematically investigates Mask R-CNN and Poin-tRend's main human video segmentation challenges, laying the groundwork for future research. This research assesses human segmentation for surveillance videos using critical discussions and future directions. It seeks to aid field researchers and practitioners.