Real Time Tobacco Leaf Classification Machine Using the Computer Vision with YOLO Approach on Jetson Nano

Closed

David F. Putradi, Budhy Setiawan, Arwin Datumaya Wahyudi Sumari

2024 Proceedings - IEIT 2024 - 2024 International Conference on Electrical and Information Technology Conference paper Cited by 3 Quartile

Abstract

This research aims at designing a tobacco leaf sorting machine that shall be employing the You Only Look Once (YOLO), specifically YOLOv5 and YOLOv8 algorithms on the Jetson Nano. The main goal is to create an efficient and automated system that can classify fresh tobacco leaves into three maturity grades. It can be classified into three main categories, which include; immature, mature, and old. The focus of this study is the mechanical design of the sorting machine, which is a central aspect of the manufacturing process. It has a 2 meters long conveyor belt and state of the art and patented flipper system to allow for high speed and accuracy in sorting. It is to be used close to the field during the post-harvest period under controlled lighting environment. Computer vision technique used will involve data cleansing, data enhancement and comparison between YOLOv5 and YOLOv8 to ascertain which of the two algorithms is best suited for this task. The YOLOv8 model proves to be more suitable for the machine due to its 85.5% accuracy with a sort time execution, 256ms, which is fast enough to control the sorting machine in real time. This sort time execution will be passed through the flipper mechanism to collect the leaves according to their maturity as stated above. The machine can perform up to 45 leaves sortation per minutes. © 2024 IEEE.

Affiliations

State Polytechnic of Malang, Malang, Indonesia