Int J Performability Eng ›› 2022, Vol. 18 ›› Issue (10): 702-709.doi: 10.23940/ijpe.22.10.p3.702-709

Previous Articles     Next Articles

Deep Learning based Aquatic and Semi Aquatic Plants Morphological Features Extraction and Classification

Jibi G. Thanikkala, Ashwani Kumar Dubeya,*, and Thomas M. T.b   

  1. aAmity School of Engineering and Technology, Amity University Uttar Pradesh, Noida, Uttar Pradesh, 201303, India;
    bDepartment of Botany, St. Thomas College, Thrissur, Kerala, 680001, India
  • Submitted on ; Revised on ; Accepted on
  • Contact: *E-mail address: dubey1ak@gmail.com
  • About author:Jibi G Thanikkal is a PhD scholar at Amity School of Engineering and Technology, Amity University Uttar Pradesh, Noida, India. Her research interests include computer vision, digital image processing and deep learning.
    Ashwani Kumar Dubey is a Professor at Amity School of Engineering and Technology, Amity University Uttar Pradesh, Noida, India. His research interests include computer vision, image processing, deep learning, smart sensors, IoT and wireless sensor networks.
    Thomas M T is an Assistant Professor of St. Thomas’ College, Thrissur, Kerala, India. His research interests include medicinal plant identification, and plant species identification.

Abstract: In Ayurveda, the ancient medicinal plant identification system is based on the morphological comparison of leaf, fruit, flower, root, stem etc. Botanists use morphometrics for aquatic and semi-aquatic medicinal plants classification. However, deep learning networks provide the highest image classification result in digital image processing. Existing deep learning algorithms generate feature maps for pixel-wise image classification. In the feature map of deep learning output, most of the morphological features are missing. This issue leads to the Catastrophic forgetting issue of deep learning. To generate a traditional morphological feature-based medicinal plant identification system, we are introducing morphometrics and morphological feature-based deep learning networks for aquatic and semi-aquatic plant classification. This article contains: (a) A detailed morphological features database of aquatic and semi-aquatic medicinal plants, (b) a summary of the importance of the morphological features-based leaf classification, (c) a morphological features extraction algorithm and (d) the morphological features-based deep learning approach for aquatic and semi-aquatic plant classification. This human brain-like procedure achieved 97% classification accuracy and reduced the Catastrophic forgetting issue of continual learning.

Key words: image processing, medicinal plants, aquatic plants, deep learning, morphometric