ARTIFICIAL INTELLIGENCE SYSTEM TO DETECT AND QUANTIFY PLASTIC BOTTLES OBJECT

• Main Author: Ramadhan Hardiansyah, Febry
• Format: Final Project
• Language: Indonesia
• Subjects: Teknik (Rekayasa, enjinering dan kegiatan berkaitan)
• Online Access: https://digilib.itb.ac.id/gdl/view/48199
• Institution: Institut Teknologi Bandung

Waste disposal has been a global problem for humankind. Pollution in our enviroment is an example caused by people disposing all kinds of trash carelessly, most notably plastic bottles. Using manual method of removal still not done efficiently factoring large areas to cover and massive amounts of human resource required. Therefore, technology is necessary to achieve solutions for the problem. One of which is the detection system technology visualized by the visualization system. The detection system technology used is YOLOv3 with transfer learning method. YOLOv3 detection system operates by using trained model which capable of determining an object. The model training is done by dataset training, which contains set of pictures that already went through image annotation process. YOLOv3 detection system provides detection result with high percentage in accuracy of 90.62%. It is approved by high detection of plastic bottle waste on the images by YOLOv3. The result of YOLOv3 detection system will be counted and visualized by using visualization system into a website. Visualization system used is Flask with HTML for the front-end and Ajax and CanvaJS for the back-end.