Purpose: This study aims to develop a classification system for the Clerodendrum paniculatum plant (Bunga Pagoda), focusing on its key parts—stems, flowers, leaves, and trees—using the Convolutional Neural Network (CNN) algorithm. The objective is to support conservation efforts and facilitate digital data grouping through technology-based classification.

Methodology: The research involved collecting a dataset of images representing different parts of the Clerodendrum paniculatum plant. These images were then used to train a CNN model. The training process included 200 epochs to optimize performance. The model's accuracy and performance were evaluated using a confusion matrix to measure classification success across the plant's various parts.

Results: The CNN model achieved its highest accuracy of 97.78% when trained for 200 epochs. The results indicated a significant improvement in evaluation metrics compared to models trained with fewer epochs. The mo   del successfully classified the plant parts with high precision, demonstrating its robustness and reliability for rare plant classification.

Conclusions: This study confirms that the CNN algorithm is effective in classifying the parts of the Clerodendrum paniculatum plant. Increasing the number of training epochs substantially enhances the model's performance, making it a practical tool for digital plant conservation initiatives.

Limitations: The study is limited by its reliance on a specific dataset, which may not encompass all possible variations of the Clerodendrum paniculatum plant under different environmental conditions.

Contributions: This research contributes to digital plant conservation by developing a CNN-based classification system for rare plants. It highlights the importance of deep learning in biodiversity preservation and provides a foundation for future AI-driven botanical studies.