“Deep Learning for Automated Crop Disease and Pest Detection: An Image-Based Diagnostic System”
A Challenge-based Computer Science master thesis project by Ling Jin, Vrije Universiteit Amsterdam, January -June 2024, supervised by Anna Bon (Vrije Universiteit Amsterdam).
As the challenges to global food security continue to increase, timely identification and management of crop diseases and pests become increasingly important. This study aims to develop a deep learning-based plant disease and pest recognition system, which can rapidly and accurately identify the types of diseases and pests affecting crops through the analysis of crop images. Utilizing an existing, crowd-sourced dataset of 20,000 labeled crop images with detailed visual characteristics of various diseases and pests, a model was trained, to achieve automatic recognition of plant diseases and pests. Additionally, a user-friendly web interface was developed to allow farmers and individuals in underdeveloped areas to upload plant photos for instant disease and pest identification, along with management suggestions. The resulting system improves identification accuracy and efficiency and enhances the feasibility and convenience of crop disease management. In this study, the process of data collection and cleaning, as well as model training and web front-end development, is detailed. Experimental results demonstrate that the model achieves satisfactory levels of accuracy in recognition, effectively supporting disease and pest management decisions, especially for resource-constrained agricultural producers. Through this research, the aim is to contribute to the sustainable development of global agricultural production and food security.