Document Type
Honors Thesis
Abstract
Darknet YOLO is an object detection software which is based around the usage of a neural network to perform its real-time predictions. YOLO stands out from other object detectors with its speed and method of prediction, illustrated in its name, “You Only Look Once,” which analyzes an image once for its predictions. Our team was tasked with training YOLO to detect objects common to an industrial warehouse setting, with these efforts soon to be followed by testing phases. Creation of the datasets involved the tracing and annotation of hundreds of images per class to be fed into YOLO’s network, which would then learn from the given input. We would utilize TinyYOLO, a lightweight version of YOLO, due to a priority in speed in training, predictions, and video feed when running YOLO. Our results were satisfactory though we would encounter some issues along the road to the final model. Work with Darknet YOLO doesnot stop with this single instance of training and is always open to a wide range of future work ideas.
Publication Date
5-1-2018
Language
English
License
This work is licensed under a Creative Commons Attribution-NonCommercial-Share Alike 4.0 International License.
Recommended Citation
Quimod, Ashley, "TRAINING DARKNET YOLO FOR USE IN AN INDUSTRIAL SETTING" (2018). 2018 Spring Honors Capstone Projects. 24.
https://mavmatrix.uta.edu/honors_spring2018/24