Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

IndoHerb: Indonesia Medicinal Plants Recognition using Transfer Learning and Deep Learning

Muhammad Salman Ikrar Musyaffa, Novanto Yudistira, Muhammad Arif Rahman, Jati Batoro

TL;DR

This study tackles the automated identification of Indonesian medicinal plants using transfer learning with CNNs. It introduces IndoHerb, a curated dataset of 10,000 images across 100 Indonesian plant classes, and evaluates five pretrained CNNs (ResNet34, DenseNet121, VGG11_bn, ConvNeXt, Swin Transformer) plus a scratch model. ConvNeXt consistently achieves the highest accuracy, reaching 92.84% on the Vietnam dataset and 92.5% on the Indonesia dataset, while the scratch model lags significantly. The results demonstrate the effectiveness of transfer learning for herbal plant recognition and underscore the potential of ConvNeXt-based approaches to support automated ethnobotanical identification and resource management. The work also highlights the value of dataset curation and augmentation in enabling robust image-classification performance for diverse plant species.

Abstract

The rich diversity of herbal plants in Indonesia holds immense potential as alternative resources for traditional healing and ethnobotanical practices. However, the dwindling recognition of herbal plants due to modernization poses a significant challenge in preserving this valuable heritage. The accurate identification of these plants is crucial for the continuity of traditional practices and the utilization of their nutritional benefits. Nevertheless, the manual identification of herbal plants remains a time-consuming task, demanding expert knowledge and meticulous examination of plant characteristics. In response, the application of computer vision emerges as a promising solution to facilitate the efficient identification of herbal plants. This research addresses the task of classifying Indonesian herbal plants through the implementation of transfer learning of Convolutional Neural Networks (CNN). To support our study, we curated an extensive dataset of herbal plant images from Indonesia with careful manual selection. Subsequently, we conducted rigorous data preprocessing, and classification utilizing transfer learning methodologies with five distinct models: ResNet, DenseNet, VGG, ConvNeXt, and Swin Transformer. Our comprehensive analysis revealed that ConvNeXt achieved the highest accuracy, standing at an impressive 92.5%. Additionally, we conducted testing using a scratch model, resulting in an accuracy of 53.9%. The experimental setup featured essential hyperparameters, including the ExponentialLR scheduler with a gamma value of 0.9, a learning rate of 0.001, the Cross-Entropy Loss function, the Adam optimizer, and a training epoch count of 50. This study's outcomes offer valuable insights and practical implications for the automated identification of Indonesian medicinal plants.

IndoHerb: Indonesia Medicinal Plants Recognition using Transfer Learning and Deep Learning

TL;DR

This study tackles the automated identification of Indonesian medicinal plants using transfer learning with CNNs. It introduces IndoHerb, a curated dataset of 10,000 images across 100 Indonesian plant classes, and evaluates five pretrained CNNs (ResNet34, DenseNet121, VGG11_bn, ConvNeXt, Swin Transformer) plus a scratch model. ConvNeXt consistently achieves the highest accuracy, reaching 92.84% on the Vietnam dataset and 92.5% on the Indonesia dataset, while the scratch model lags significantly. The results demonstrate the effectiveness of transfer learning for herbal plant recognition and underscore the potential of ConvNeXt-based approaches to support automated ethnobotanical identification and resource management. The work also highlights the value of dataset curation and augmentation in enabling robust image-classification performance for diverse plant species.

Abstract

The rich diversity of herbal plants in Indonesia holds immense potential as alternative resources for traditional healing and ethnobotanical practices. However, the dwindling recognition of herbal plants due to modernization poses a significant challenge in preserving this valuable heritage. The accurate identification of these plants is crucial for the continuity of traditional practices and the utilization of their nutritional benefits. Nevertheless, the manual identification of herbal plants remains a time-consuming task, demanding expert knowledge and meticulous examination of plant characteristics. In response, the application of computer vision emerges as a promising solution to facilitate the efficient identification of herbal plants. This research addresses the task of classifying Indonesian herbal plants through the implementation of transfer learning of Convolutional Neural Networks (CNN). To support our study, we curated an extensive dataset of herbal plant images from Indonesia with careful manual selection. Subsequently, we conducted rigorous data preprocessing, and classification utilizing transfer learning methodologies with five distinct models: ResNet, DenseNet, VGG, ConvNeXt, and Swin Transformer. Our comprehensive analysis revealed that ConvNeXt achieved the highest accuracy, standing at an impressive 92.5%. Additionally, we conducted testing using a scratch model, resulting in an accuracy of 53.9%. The experimental setup featured essential hyperparameters, including the ExponentialLR scheduler with a gamma value of 0.9, a learning rate of 0.001, the Cross-Entropy Loss function, the Adam optimizer, and a training epoch count of 50. This study's outcomes offer valuable insights and practical implications for the automated identification of Indonesian medicinal plants.
Paper Structure (25 sections, 27 figures, 4 tables)

This paper contains 25 sections, 27 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. IndoHerb Dataset
  4. Methodology
  5. Gathering Data
  6. Preprocessing Data
  7. Structuring Model
  8. Testing Model
  9. Experiments
  10. Testing the Vietnam Medicinal Plant Dataset
  11. Testing Vietnam Dataset with pre-trained model ResNet34
  12. Testing Vietnam Dataset with pre-trained model DenseNet121
  13. Testing Vietnam Dataset with pre-trained model VGG_bn
  14. Testing Vietnam Dataset with pre-trained model ConvNeXt_base
  15. Testing Vietnam Dataset with pre-trained model Swin_t
  16. ...and 10 more sections

Figures (27)

  • Figure 1: Graph of Indonesia Medicinal Plant Dataset Distribution Before Augmentation
  • Figure 2: Total Images Before and After Augmentation
  • Figure 3: Example Images From Five Selected Classes of Indonesia Medicinal Plant Dataset
  • Figure 4: Graph of ResNet34 Model Loss Value from Vietnam Dataset Against Epoch
  • Figure 5: Graph of ResNet34 Model Accuracy Value from Vietnam Dataset Against Epoch
  • ...and 22 more figures