BLOB DETECTION IN STATIC CAMERA FOR HUMAN COUNTING APPLICATION

• Main Author: YOGA ESA WIBOWO (NIM: 13214012), FRANSISKUS
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/27397
• Institution: Institut Teknologi Bandung

<p align="justify"> Automatic human counting is applicable in various application including the enhancement of building security sistem. For such application, the static camera has several advancements such as the potency in the implementation of less complex algorithm. Visitor counter, human counter and validator device, is presented as altenative solution for that problem. This document specially discussed the design and implementation of object detection process in visitor counter device. The input of this process is realtime camera feed and the outputs are detected object location and its bounding box. Detection process is devided into 3 step which are background segmentation, contour detection and sub-sampling, and clustering. In the clustering process itself, it uses Gaussian Mixture Model with expectation minimization approach and silhouette index evaluation. According to the test result, the proposed algorithm is able to recognize contour with 2 object walking side-by-side and fail to recognize 2 object walking in a queue (back-and-front). <p align="justify">