<![CDATA[Ornamental plants are widely used in residential and public environments due to their aesthetic value, yet some species contain toxic compounds that may pose risks to humans and animals. The visual similarity between toxic and non-toxic ornamental plants makes manual identification difficult for the general public, increasing the potential for accidental exposure. This study proposes an automated image-based classification system to identify toxic and non-toxic ornamental plants accurately and efficiently. The system utilizes deep learning-based image classification techniques to analyze plant images and categorize them into five classes: lily, daffodil, caladium, sansevieria, and non-toxic ornamental plants. A dataset of 3,320 images was prepared and divided into training, validation, and testing subsets. Image preprocessing and augmentation were applied to improve data quality and model generalization. Experimental results show that the proposed model achieved a testing accuracy of 98.25%, outperforming the comparative model and demonstrating stable classification performance across all classes. The best-performing model was integrated into a web-based application that enables users to upload plant images and receive real-time identification results. The findings indicate that the proposed system provides an accurate, accessible, and practical solution for identifying toxic ornamental plants and supporting public safety awareness. Future development may focus on expanding plant categories and deploying mobile-based implementations for broader accessibility.]]>
Copyrights © 2026
