Crop Classification System in Agriculture

Izanoordina Ahmad; Syed Harith Aidid Aljamalullil   Syed Khair Azmir Jamalulil; Nor Hidayah   Abdul Kahar; Noorazlina   Mohamid Salih; Danial   Md Noor; Indrarini Dyah   Irawati

doi:10.37934/araset.63.2.201212

Authors

Izanoordina Ahmad Electronics Technology Section, Universiti Kuala Lumpur British Malaysian Institute, 53100 Gombak, Selangor, Malaysia
Syed Harith Aidid Aljamalullil Syed Khair Azmir Jamalulil Electronics Technology Section, Universiti Kuala Lumpur British Malaysian Institute, 53100 Gombak, Selangor, Malaysia
Nor Hidayah Abdul Kahar Electrical Engineering Section, Universiti Kuala Lumpur British Malaysian Institute, 53100 Gombak, Selangor, Malaysia
Noorazlina Mohamid Salih Marine and Electrical Engineering Technology Section, Universiti Kuala Lumpur Malaysian Institute of Marine Engineering Technology (UniKL-MIMET), 32200 Lumut, Perak, Malaysia
Danial Md Noor Faculty of Electrical and Electronic Engineering, Universiti Tun Hussein Onn Malaysia, Parit Raja, 86400 Batu Pahat, Johor, Malaysia
Indrarini Dyah Irawati School of Electrical Engineering, Telkom University, Kabupaten Bandung, Jawa Barat 40257, Indonesia

DOI:

https://doi.org/10.37934/araset.63.2.201212

Keywords:

YOLOv8, Computer vision, Roboflow, Convolutional neural network (CNN), Ultralytics

Abstract

Agriculture plays a crucial role in Malaysia's economy, providing employment opportunities, income, and ensuring food security. The country's diverse climate and geography contribute to a wide range of agricultural products. However, farmers face challenges in identifying crop diseases and damage, leading to time‐consuming inspections and potential human errors during large‐scale crop harvesting. In order to address these issues, this project aims to design and develop a computer vision‐based system for classifying the condition of crops. By leveraging computer vision, the system eliminates the need for manual inspection and empowers farmers with an efficient and accurate tool for assessing crop quality. Visual cues, such as colour are automatically analysed by the YOLOv8 model to detect signs of damage in each chili crop. The crop classification system was evaluated using a separate dataset consisting of 300 chili samples. The results demonstrated the effectiveness of the YOLOv8 model, achieving an impressive accuracy of 99.67%. The system exhibited perfect precision in identifying bad chilis and a recall rate of 99.33% for capturing the majority of bad chilis present in the dataset. This study signifies the potential of the YOLOv8 model as a reliable tool for predicting chili crop quality. By implementing this system, farmers can save significant time and effort previously spent on manual inspections.