Elderly Hajj pilgrims activity recognition based on candidate classification technique

Abstract

Every year, the rate of elderly people is rising, and monitoring them in mass gatherings like a pilgrimage is a tough task. Millions of Muslims around the world visit Makkah and Madinah for the Hajj pilgrimage. Since 1990, more than 4800 pilgrims have died in the stampede. Aged pilgrims are very much vulnerable to these stampedes. To support them, it is very important to monitor their activities. But only a little is done to monitor the elderly pilgrims. In this article, all the rituals performed in the Hajj pilgrimage have been studied and an activity recognition dataset for elderly pilgrims in line with the rituals of Hajj has been created. A novel classification method called the candidate classification technique (CCT) is proposed to recognize the activities accurately with less response time. Three publicly available activity recognition datasets namely UCI HAR, USC HAD, SKODA, and the proposed elderly hajj pilgrim activity recognition dataset have been used. The accuracy achieved using the proposed classification technique is 98.25%, 96.97%, and 99.29% for the UCI HAR, USC HAD, and SKODA datasets.