Image dataset on the Chinese medicinal blossoms for classification through convolutional neural network

Area of Science:

• Botany
• Pharmacology
• Computer Science

Background:

• Traditional Chinese medicine extensively utilizes tree blossoms for disease prevention and treatment.
• Edible flowers offer nutritional, medicinal, cosmetic, and aromatic benefits, enhancing gourmet food.
• Accurate identification of edible flowers is critical due to visual similarities and potential toxicity.

Purpose of the Study:

• To address the lack of a dedicated dataset for Chinese medicinal blossoms.
• To present and establish a dataset of twelve common and economically valuable blossoms used in traditional Chinese medicine.
• To support the accurate classification of traditional Chinese herbs by pharmacists and researchers.

Main Methods:

• Compilation of a novel image dataset featuring twelve key Chinese medicinal blossoms.
• Dataset designed to facilitate image segmentation and classification tasks.
• Resource intended for machine learning and deep learning algorithm development.

Main Results:

• Establishment of a comprehensive dataset for commonly used Chinese medicinal blossoms.
• The dataset aids in distinguishing between visually similar flower species.
• Provides a foundation for automated classification systems.

Conclusions:

• The developed blossom dataset is vital for traditional Chinese medicine applications.
• It enhances the accuracy and efficiency of herb identification.
• Serves as a valuable resource for advancing machine learning in pharmacognosy.