Date of Graduation


Document Type


Degree Name

Bachelor of Science


Computer Science and Computer Engineering


Luu, Khoa

Committee Member/Reader

Nelson, Alexander

Committee Member/Second Reader

Gauch, John


The Insect Detection Server was developed to explore the deployment and integration of an Artificial Intelligence model for Computer Vision in the context of insect detection. The model was developed to accurately identify insects from images taken by camera systems installed on farms. The goal is to integrate the model into an easily accessible, cloud-based application that allows farmers to analyze automatically uploaded images containing groups of insects found on their farms. The application returns the bounding boxes and the detected classes of insects whenever an image is captured on-site, enabling farmers to take appropriate actions to address the issue of the insects' presence. To extend the capabilities of the application, the server is linked to a GPT-3.5 API. This will allow users to ask questions about the bugs detected on their farms, creating an "online expert"-like feature. Python, C++, and Computer Vision libraries were used for the detection model, while the OpenAI API was used for GPT-3.5's integration. By combining these technologies, farmers can more effectively and efficiently manage pests on their farms than current alternatives. This Generative Pre-trained Transformer (GPT) aspect of the project can be leveraged to enable the emulation of agricultural experts for users/farmers. The large language model (LLM) neural network can be fine-tuned using prompt engineering to generate natural language responses to user queries. This will enable farmers to get expert advice and guidance on pest management without having to consult with a human expert. The integration of GPT-3.5 API will also allow the application to provide personalized recommendations based on each farm's specific needs and circumstances. This added feature will give the farmers a more comprehensive and tailored approach to pest management, further increasing the efficiency and effectiveness of their pest control strategies. The significance of this research lies in the development of a practical and accessible tool for farmers to manage pests on their farms. Using Computer Vision and Artificial Intelligence, farmers can quickly and accurately identify insects, leading to more efficient and effective pest management. This could help farmers reduce the use of pesticides and other forms of pest management, leading to improved crop yields and reduced environmental impacts. The potential benefits of this technology extend beyond the agricultural industry, as the techniques used in this research can be applied to a wide range of computer vision and user-facing data analytic applications. For example, the developed techniques could be applied to other fields, such as surveillance, security, and medical imaging.


Artificial Intelligence, Agriculture, Transformers, ChatGPT, Cloud Deployment, Large Language Models