Abstract

Mangroves are unique coastal ecosystems that are rich in biodiversity and have significant ecological value. Identifying mangrove species is important for many applications, such as biodiversity, restoration, and monitoring. As traditional methods are complicated and time-consuming, non-experts need an approach to identify mangroves in a timely and cost-effective manner. In this study, we created a deep learning approach for mangrove species identification based on leaf image recognition. We used digital images of mangrove leaves to identify mangrove species by applying Convolutional Neural Networks (CNN). A dataset of leaf images from 11 ‘true’ mangrove species found in Bali, Indonesia, was developed and divided into 80% for training and 20% for test datasets. About 20% of the training dataset was used for validation. Our results showed an accuracy of 98.86% on validation and 97.16% on a test set of images, promising possibilities for mangrove species identification. The finding indicates that the model effectively identifies mangrove species that are high in diversity and have morphological similarities.

Keyword: Bali, Convolutional Neural Networks, deep learning, Indonesia, mangroves, plant identification