Theses and Dissertations
ORCID
https://orcid.org/0009-0009-5328-2850
Issuing Body
Mississippi State University
Advisor
To, Filip Suminto D.
Committee Member
Zhang, Xin
Committee Member
Chen, Jingdao
Date of Degree
12-13-2024
Original embargo terms
Complete embargo 1 year
Document Type
Graduate Thesis - Open Access
Major
Engineering (Biosystems Engineering)
Degree Name
Master of Science (M.S.)
College
James Worth Bagley College of Engineering
Department
Department of Agricultural and Biological Engineering
Abstract
The thesis presents an application of a convolutional neural network (CNN) to detect and classify plastic contaminants in seed cotton in real-time, in a ginning environment. A multi-layered CNN was developed and used to detect plastic contaminants from images captured by a universal serial bus (USB) camera, in a flowing cotton stream, with an artificial lighting system. The CNN was trained using a collection of images captured in the dynamic setting that simulated real-time cotton flow using a custom design feeder system. The CNN was trained in Google Colaboratory (Colab) environment. After the training was completed, it was then converted into TensorRT format and loaded into a Jetson Xavier-based embedded system. The model was able to achieve an 87-93% detection accuracy while maintaining 30-50 frames per second (FPS).
Recommended Citation
Harjono, Jonathan, "Plastic contaminants detection and classification in seed cotton using customed design feeder separator unit and convolutional neural network" (2024). Theses and Dissertations. 6386.
https://scholarsjunction.msstate.edu/td/6386