Klasifikasi Tanaman Air Berdasarkan Kebutuhan Cahaya pada Aquascape Menggunakan Algoritma YOLO

Abstract

Aquatic plants are plants that live in aquatic environments. These plants are often used as components to decorate aquascapes because of their attractive appearance and forms. Aquascaping is an art that involves combining aquatic plants, ornamental fish, wood, rocks, and other decorations within an aquarium to create a beautifully integrated natural underwater ecosystem. Aquatic plants require light to grow well, and each type has different light requirements. Therefore, a system was developed to classify images of aquatic plants. Several methods can be used for image classification in this field, one of which is the YOLO algorithm. First, data collection was conducted, gathering 226 images of aquatic plants categorized by their light requirements. Next, image preprocessing was performed using the Roboflow platform. Roboflow is capable of data preprocessing, from image data splitting to augmentation. The result of preprocessing on the Roboflow platform was a dataset ready for training. Training was carried out using Google Colab Notebooks with the YOLO method. To find the best model, fine-tuning of the hyperparameters was also performed. The best-trained model was then used in the developed system to classify aquatic plants. Test results showed that the YOLO algorithm could classify the light requirements of aquatic plants with quite high accuracy, achieving 80.95%. This method can assist aquascape enthusiasts in selecting aquatic plants based on their light needs.