Deep learning and drone imagery-based automated recognition of coffee plant varieties | |
| Author | Wangmo, Chime |
| Call Number | AIT Thesis no.AE-24-03 |
| Subject(s) | Coffee Agricultural innovations Drone aircraft Neural networks (Neurobiology) |
| Note | A Thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Agricultural Systems and Engineering |
| Publisher | Asian Institute of Technology |
| Series Statement | Thesis ; no. AE-24-03 |
| Abstract | Coffee, a globally traded agricultural commodity and one of the most consumed beverages worldwide, plays a significant role in generating millions of jobs and income. The expansion of the coffee industry driven by increased consumption and demand for specialty coffee necessitates innovative methods for accurately identifying and classifying coffee plant varieties. Traditional approaches based on physical characteristics and chemometric techniques face challenges as the number of coffee varieties grows. The study aims to investigate the viability of utilizing images captured by unmanned aerial vehicles (UAVs) with high-resolution sensors for images processing techniques. This involves the collection, processing, and analysis of real-time drone imagery to identify different coffee plant varieties. Moreover, the research focuses on the development and optimization of deep learning algorithms, particularly Convolutional Neural Networks (CNNs), tailored specifically for discerning unique features of various coffee plant varieties. CNNs belong to deep neural networks that have proven highly effective in various computer vision tasks, such as object detection, image classification and image segmentation. By integrating advanced technologies such as UAVs and CNNs, the study seeks to enhance the efficiency and accuracy of coffee variety classification, offering potential advancements for the coffee cultivation sector. The CNNs model achieved an 89.1% training accuracy, 67.8% validation accuracy, and an overall predictive accuracy of 67.06%. These results demonstrate the effectiveness of employing advanced computer vision techniques for coffee variety classification using drone-captured imagery. This study offers a promising decision support algorithm for the classification of coffee plant varieties, contributing to the enhancement of the coffee cultivation sector. By revolutionizing the coffee industry through technologically advanced and efficient solutions, this research addresses the challenges associated with the growing diversity of coffee varieties. |
| Year | 2024 |
| Corresponding Series Added Entry | Asian Institute of Technology. Thesis ; no. AE-24-03 |
| Type | Thesis |
| School | School of Environment, Resources, and Development |
| Department | Department of Food, Agriculture and Natural Resources (Former title: Department of Food Agriculture, and BioResources (DFAB)) |
| Academic Program/FoS | Agricultural Systems and Engineering (ASE) |
| Chairperson(s) | Himanshu, Sushil Kumar; |
| Examination Committee(s) | Datta, Avishek;Pramanik, Malay; |
| Scholarship Donor(s) | Asian Development Bank-Japan Scholarship Program (ADB-JSP); |
| Degree | Thesis (M. Eng.) - Asian Institute of Technology, 2024 |