Description
Abstract:This study aims to ascertain the viability of using a machine learning network to detect sunburn on grapes. This will be done via experiments done using a deep learning network created using the tool MATLAB. The network will classify the sunburn as: none, light, medium or heavy sunburn. On a larger scale this has the potential to provide useful data regarding grape yields and may highlight problem areas. The primary goal of this network is the capability to identify grape sunburn at an accuracy of at least 50% with an optimal goal of 75%. The primary aim of this level of accuracy is to test the viability of the network rather than make a commercial use product. An accuracy of 50% or higher means the network is more often right than wrong. An accurate detection network provides opportunity to further studies in the field of grape sunburn by allowing important statistics to be gathered on the area. These statistics could be of great use in assisting in mitigation or avoidance strategies for sunburn or even for gathering more data on the causes of the phenomenon.
| Period | 01 Feb 2021 → 10 Dec 2021 |
|---|---|
| Examinee | Isaac Prior |
| Examination held at | |
| Degree of Recognition | International |