Castillon, Joshua A.  <br/><br/>
Data-driven mobile application for native trees using ai-identification/ 
Joshua A. Castillon, Cristopher S. Chan, Alliana Hira G. Cordero, Ken Vaness M. Fundales, and Angel Junine P. Peñalosa.-- 
 - Manila:  Technological University of the Philippines,  2025. 
 - xxiii, 197pages:  29cm. 
<br/><br/> Bachelor's thesis 
<br/><br/>College of Industrial Technology.-- 
<br/><br/>Includes bibliographic references and index. 
<br/><br/> The increasing loss of native tree species in urban environments, especially in<br/>Metro Manila, underscores the urgent need for products that can provide new means to<br/>access tools to raise awareness and identify and protect local flora. This study describes the<br/>creation of a data-driven mobile application to identify five native trees from the<br/><br/>Philippines - Narra, Kamagong, Talisay, Banaba, and Ipil - using artificial intelligence-<br/>based leaf image identification. Using the mobile application, users can accurately identify<br/><br/>trees and receive information on each tree species, including scientific names, ecological<br/>preferences, and growth habits. The system was equipped with an intuitive interface that<br/>includes a searchable tree database, as well as a map function to showcase important areas<br/>in the Philippines with each tree species. Nevertheless, the identification model was<br/>specifically trained only using native trees in and around the Arroceros Forest Park region,<br/>thus limiting appropriate identification to only those species in that park. The AI<br/>identification model includes a deep learning convolutional neural networks (CNN) trained<br/>on a designated leaf image dataset that was curated, enhancing performance reliability. The<br/>application was tested against a number of devices and Android versions, and the results<br/>showed excellent functioning, responsive controls, and consistent behavior. Limitations<br/>were the limitation to five tree species, and the identification accuracy was only relevant<br/>to those in Arroceros. Testing was done against the software quality standards ISO/IEC<br/>25010 and showed that the system’s accuracy, functionality, and usability were adequate.<br/>This project is the start of an initiative for urban biodiversity by providing an object-based<br/>and usable tool for native tree identification, as well as advocacy for tree conservation. 
<br/><br/><label>Subjects--Topical Terms: </label><br/>Mobile application<br/>Conservation advocacy<br/>Image recognition
<br/><br/><label>LC Class. No.: </label>BTH QA 76.9  / C37 2025