IMPLEMENTATION OF YOLOV4 OBJECT DETECTION ALGORITHM ON THERMAL INFRARED OBJECT TRACKING SYSTEM

• Main Author: Ragitya Anggaraksa, Fauzan
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/56345
• Institution: Institut Teknologi Bandung

In this final project, a human object detection system was designed using an infrared thermal camera. The detection is carried out from a height of ±5 meters. The system is designed using a camera holder that is connected to a servo so that the camera can track objects detected by the image servoing method. This system consists of user interface subsystem, data processing subsystem, acquisition subsystem, and driver subsystem. The user interface subsystem is a subsystem that has a function to connect the user and the device. The data acquisition subsystem is a subsystem consisting of infrared camera components and has a function to capture thermal information from objects. The data processing subsystem is a subsystem consisting of the NVIDIA Jetson Nano Mini PC component and has functions as the brain of the system as a whole and plays a role in running object detection features with the YOLOv4 algorithm along with sending control signals to the drive subsystem to move the camera in the acquisition subsystem. The drive subsystem is a subsystem consisting of two servos to drive the infrared camera platform so that the object being tracked is always in the frame. The main topic in this report is the process of implementing the YOLOv4 object detection algorithm in the Thermal Infrared Object Tracking System.