Behavioural Analysis For Prospects In Crowd Emotion Sensing: A Survey

Crowd behavioural analysis is an interesting and emerging domain in research, with incomplete set of activities, tasks and lack of intermediate cub-processes which are mandated for productive analysis. Since the domain is untapped to a major extent, the research carried out in the domain needs proper stages of operations. A proper taxonomy will direct the futuristic domains in the right track of processes and organization of intermediate tasks. This review paper intends to document the list of stages and processes, data collection, pipelining the sub-tasks, pre-emptive identification of supposed problems during the later stages in detection of crowd emotions and behavioural analysis. Deep learning techniques have been widely deployed to investigate the models of crowd analysis, anomaly detection, and look for meaningful insights and patterns from datasets. The Different models are investigated thoroughly for their respective understanding about the emotional aspects considered in the studies. Emotional characteristics when powered with crowd behavioural analysis and real world entities will deliver a promising solution for crime detections, anomaly detection and ensure a safer environment for nations. Video surveillance tools, datasets from crime datasets and various other factors contributed to the previous research works, models are now being designed to incorporate the best features of these models into one and thus achieve one fruitful model for continuous video analytics.