Incremental and Zero-Shot Machine Learning for Vietnamese Medicinal Plant Image Classification - KES 2024

Aug 5, 2026

Upload by Eden

This paper studies incremental learning and zero-shot learning for classifying Vietnamese medicinal plants from images, addressing the challenge that new species continually emerge and appearances vary. The method combines incremental learning to update the model with new data while preserving prior knowledge, and zero-shot learning to recognize unseen species via semantic similarity. Experiments on a Vietnam-specific dataset demonstrate improved adaptability and robustness, with reduced reliance on extensive labeled data, supporting dynamic medicinal plant identification scenarios.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

AccessFree to download

Introduction

Context and Importance

Research Gap

Research question

Related studies

Incremental learning

[Available online at](Available online at www.sciencedirect.com)[ www.sciencedirect.com](Available online at www.sciencedirect.com) ScienceDirect Procedia Computer Science 246 (2024) 606–615 28th International Conference on Knowledge-Based and Intelligent Information & Engineering Systems (KES 2024) Incremental and Zero-Shot Machine Learning for Vietnamese Medicinal Plant Image Classification Trien Phat Trana,∗, Fareed Ud Dina, Ljiljana Brankovica, Cesar Saninb, Susan M Hestera,c, Minh Duc Hoang Led a The University of New England, Armidale NSW 2351, Australia b The University of Newcastle, Callaghan NSW 2308, Australia c The University of Melbourne, Parkville VIC 3052, Australia d Thanh Dong University, Hai Duong, Vietnam Abstract This paper presents a study on the use of incremental and zero-shot learning for classifying Vietnamese medicinal plants using image analysis. Traditional machine learning methods often struggle with the constant emergence of new plant species and variability in appearances. Our methodology combines incremental learning, which continuously updates the model with new data while retaining prior knowledge, and zero-shot learning, which classifies unseen plant species by leveraging semantic similarities. Evaluated on a unique dataset from Vietnam, our approach shows improved adaptability, robustness, and reduced dependency on extensive labeled data, making it suitable for dynamic environments like medicinal plant identification. © 2024 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license ([https://creativecommons.org/licenses/by-nc-nd/4.0](https://creativecommons.org/licenses/by-nc-nd/4.0)) Peer-review under responsibility of the scientific committee of the 28th International Conference on Knowledge Based and Intelligent information and Engineering Systems Keywords: incremental machine learning; zero-shot learning; image classification, medicinal plants; openai clip 1. Introduction 1.1. Context and Importance The field of image classification has witnessed remarkable advancements with the advent of deep learning techniques, such as convolutional neural networks (CNNs) [1] . State-of-the-art architectures like VGGNet [2], ResNet [3], and EfficientNet [4] have achieved impressive performance on various benchmarks. However, these models often require extensive labeled training data and struggle to adapt to new or evolving classes without retraining from scratch. ∗ Corresponding author. Tel.: +61-404-645-878 . E-mail address: [ttran72@myune.edu.au](ttran72@myune.edu.au) 1877-0509 © 2024 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license ([https://creativecommons.org/licenses/by-nc-nd/4.0](https://creativecommons.org/licenses/by-nc-nd/4.0)) Peer-review under responsibility of the scientific committee of the 28th International Conference on Knowledge Based and Intelligent information and Engineering Systems 10.1016/j.procs.2024.09.469 Trien Phat Tran et al. / Procedia Computer Science 246 (2024) 606–615 607 This limitation presents significant challenges in dynamic environments like medicinal plant identification, where new species or variations may emerge continually. Incremental learning, also known as continual learning, offers a promising solution by enabling models to acquire new information from a changing data stream while retaining previously learned knowledge [5] . This capability is crucial in agricultural settings where new plant species or diseases can emerge over time. Furthermore, zero-shot learning (ZSL) techniques allow models to classify instances into classes that were not seen during the training phase [6, 7], further enhancing their adaptability and reducing the need for extensive data labeling. 1.2. Research Gap There is a notable gap in the application of incremental and zero-shot learning for the classification of medicinal and h

FAQs

Why are incremental learning and zero-shot learning needed for medicinal plant image classification?

Traditional models require many labeled samples and struggle when new plant species appear or when visual appearances vary. Incremental learning updates with new data while retaining earlier knowledge, and zero-shot learning can classify unseen species using semantic similarity.

How does the proposed approach combine incremental and zero-shot learning?

The approach uses incremental learning to continuously integrate new images into the model without forgetting previously learned classes, and zero-shot learning to infer classes not present in training by leveraging semantic relationships.

What evidence supports the effectiveness of the method?

The paper evaluates the approach on a unique Vietnam dataset and reports improved adaptability, robustness, and less dependence on large labeled datasets, making it suitable for dynamic environments like medicinal plant identification.

Aug 5, 2026

Upload by Eden