Repository logo
 

YOLO Models for Fresh Fruit Classification from Digital Videos

Files

Thesis(3.45 MB)

Date

2023

Authors

Supervisor

Yan, Weiqi

Item type

Thesis

Degree name

Master of Computer and Information Sciences

Journal Title

Journal ISSN

Volume Title

Publisher

Auckland University of Technology

Abstract

With the development of deep learning and computer vision, daily activities are rapidly being replaced by computers or robots, for example, unmanned vehicles. Recently, with the advanced Microsoft Copilot, even coding work can be replaced by machines, there’s no doubt that our lives become more and more convenient based on that technological development. In this thesis, we conduct research work on fruit freshness detection from digital images. In the first step, we take use of YOLOv6, YOLOv7, and YOLOv8 to detect a variety of fruits from digital images, this can incredibly improve the efficiency and accuracy of fruit classification compared with human work. To generate the final outcomes, we train the three architectures individually, and we adopt a majority vote to get a better performance for fresh fruit detection. Compared with the previous work, our clustering method has higher accuracy and is faster than the previous architectures. Because we are use of the clustering method to generate our final results, it will be easy for us to change the backbone and get a better result in the future. Pertaining to the dataset, we selected a dataset from Kaggle which includes 6 classes for the fruits, namely, “fresh apple”, “fresh banana”, “fresh orange”, “rotten apple”, “rotten banana”, and “rotten orange”. Based on this dataset, the proposed deep learning models are able to detect visual objects from digital images taken by ourselves, which can be easily fine-tuned with a larger dataset in the near future.

Description

Keywords

Freshness detection, YOLOv6, YOLOv7, YOLOv8, orange, apple, banana

Source

DOI

Publisher's version

Rights statement

Permanent link

http://hdl.handle.net/10292/17702

Collections

Masters Theses