Pengenalan Aktivitas Manusia pada Area Tambak Udang dengan Convolutional Neural Network

Abstract

Thievery is a problem that can harm theft victims. Thievery usually occurs at night when there is no supervision of goods in a location. To avoid thievery and monitor conditions in a location, CCTV (Closed-Circuit Television) cameras can be used. However, the function of CCTV camera systems is only a passive monitoring systems. In this paper, a human activity recognition is designed using CCTV cameras to produce a security system. Inputs on the recognition process are videos obtained from CCTV cameras installed in the shrimp pond. Human activity recognition that is used in this study is Convolutional Neural Network. Before the human activity recognition was carried out, the program first detected humans with the YOLO (You Only Look Once) algorithm and tracking it with the SORT (Simple Online and Realtime Tracking) algorithm. The results obtained from the human activity recognition is class labels on human objects that are tracked.