<![CDATA[This study aims to develop and evaluate a deep learning-based mobile application for the automated classification of ornamental plants, addressing challenges associated with visual similarity among species and environmental variability during image acquisition. The proposed system utilizes a Convolutional Neural Network (CNN) based on the MobileNetV2 architecture, selected for its lightweight structure and deployment efficiency on resource-constrained mobile devices. The dataset comprises approximately 600 images representing 10 ornamental plant classes, collected from real-world environments, and processed through a standardized preprocessing pipeline. Model training was conducted using the Teachable Machine platform over 100 epochs, with a batch size of 16 and a learning rate of 0.001, allocating 90% of the dataset for training and 10% for testing. Experimental results indicate that the proposed model achieves a classification accuracy of 96.3%, corroborated by evaluation metrics including accuracy curves, loss convergence, and class-wise performance analysis. The trained model was successfully converted into a lightweight format and integrated into an Android-based mobile application developed using the Flutter framework. Functional testing demonstrates that the application performs effectively in real-time classification scenarios, maintaining high accuracy and responsive on-device inference without relying on cloud computing. In conclusion, this study confirms that lightweight deep learning architectures, such as MobileNetV2, can be effectively implemented in mobile environments for ornamental plant classification. The proposed application enhances accessibility and usability, enabling rapid and accurate plant identification. Furthermore, this approach contributes to practical applications in horticulture, education, and biodiversity awareness, while demonstrating the feasibility of deploying efficient deep learning models on mobile platforms.]]>
Copyrights © 2026
