Andi Baso Kaswar, Yasser Abd Djawad, Oslan Jumadi, Abd. Muis, Ridwansyah
The use of rain shelters is intended to prevent corn plant flowers from being miscarried because of high rainfall. However, the effects of rain shelters on corn growth have not been explored. Therefore, this study analyzed the growth of maize using rain shelters based on leaf color. One-way farmers can identify corn fertility by making visual observations using leaf color charts. However, this method is prone to misinterpretation owing to farmer subjectivity. Several studies have developed various methods to overcome these problems; however, these methods require expensive and complex equipment and specialized user competence. Therefore, this study proposes a classification system for corn plant fertility based on leaf images using the CNN ResNet50 architecture with an image enhancement process. The proposed method uses a smartphone camera with additional microlens during the image-acquisition process. The proposed method can provide a loss value of 15.16%, accuracy of 95.35%, precision of 95.35%, recall of 95.25%, AUC of 98.48%, and F1-Score of 95.33%; the computation time of the training process is 1075.94 seconds. The test results demonstrated that the proposed method can accurately classify the treatment of corn plants using rain shelter. © 2025, Chulalongkorn University, Faculty of Fine and Applied Arts. All rights reserved.
Computer Engineering Department, Universitas Negeri Makassar, Indonesia; Electronics Engineering Department, Universitas Negeri Makassar, Indonesia; Biology Department, Universitas Negeri Makassar, Indonesia
