DEVELOPMENT OF AN INTELLIGENT PLANT PEST DETECTION ALGORITHM USING TRANSFER LEARNING TECHNIQUE

Read full text Summarise with VIRAT
Cited by: 0
References: 12

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The rising threat of plant pests acts as a major threat in case of agricultural productivity whose consequent outcome is reduced crop yield and economic losses.This paper shows the creation of an intelligent plant pest detection system, considering a deep learning solution through the ResNet50 network architecture of convolutional neural network.The model was trained and tested using a single publicly available dataset of pest images of 6 major pest categories such as Aphid, Armyworm, Whitefly, Mealybug, Thrips and Leafhopper.Preprocessing and augmentation of the dataset were also done to increase the learning ability of the model.ResNet50 was modified to be an efficient distributor of features and classifier using transfer learning.The model attained a high rate of average accuracy 91.63% with 91.01% and 90.35% on average precision and F1score, respectively.The performance of ResNet50 stood out when a compare and contrast analysis with other architectures was conducted, i.e.VGG16 and MobileNet.The suggested system has an excellent potential to be implemented in smart agriculture and allows to determine the presence of pests in real-time and with proper accuracy, allowing to employ early intervention measures and enhance the health management of crops.Additional areas of work will be in the larger range of pests and the system integration into mobile application system which will be useful at the field level.

Fields of Study & Subjects

Smart Agriculture and AI Advanced Neural Network Applications Industrial Vision Systems and Defect Detection

Publication details

Journal / source
International Research Journal of Modernization in Engineering Technology and Science
Publication date
25 March 2026
Language
EN