DESIGN AND IMPLEMENTATION OF AI SYSTEMS FOR OBJECT DETECTION (CASE STUDIES PERSON OBJECT)

• Main Author: Aulia Padmanagara, Muthahhari
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/39937
• Institution: Institut Teknologi Bandung

Nowadays the development of AI (Artificial Intelligence) is very rapid and covers various disciplines. For example, an AI system is expected to be able to obtain important information from a text, an image, and any others data. One of the uses of this AI system is to be able to detect objects from the given image. Of course, this is very necessary for various things, let's say for security, self-driving car, etc. In this Final Project an AI system will be developed which aims to be able to detect person class objects, this system is expected to be reliable, which has a high value of accuracy and speed. The process of forming a system uses a reference from a system that is considered to be capable of detecting objects and making adjustments to the system to fit in with case studies. The system is in the form of a transfer learning method that is the training based on a system that already exists before, this training is concluded so that the system can specifically detect objects with the person class. The system that has been built will then be evaluated by looking at the value of accuracy and speed of the system. This evaluation uses both metrics suggested by the dataset provider team. These results will then be compared with several similar object detection system methods. The output of this study is expected to form an AI object detection system that reliable for detecting person class.