Innovative Insect Detection and Classification for the Agricultural Sector Using Gannet Optimization Algorithm With Deep Learning

Generally, the agricultural field can enhance food needs and deliver healthy and nutritious food. Recognizing and classifying crop insects is a crucial threat for agriculturalists as it is essential in preventing crop damage and preserving quality. Traditional pest recognition models often needed mo...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE access Jg. 12; S. 108041 - 108051
Hauptverfasser: Al-Shahari, Eman A., Aldehim, Ghadah, Sharaf Almalki, Nabil, Assiri, Mohammed, Sayed, Ahmed, Alnfiai, Mrim M.
Format: Journal Article
Sprache:Englisch
Veröffentlicht: IEEE 2024
Schlagworte:
Adaptation models
Agriculture
Biological system modeling
computer vision
Crops
deep learning
Feature extraction
gannet optimization algorithm
hyperparameter tuning
insect detection
Insects
Optimization
Tuning
ISSN:2169-3536, 2169-3536
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Generally, the agricultural field can enhance food needs and deliver healthy and nutritious food. Recognizing and classifying crop insects is a crucial threat for agriculturalists as it is essential in preventing crop damage and preserving quality. Traditional pest recognition models often needed more knowledgeable taxonomists who could precisely detect pests based on morphological aspects. Pest detection employing deep learning (DL) is a respected use of artificial intelligence (AI) in fields such as entomology, agriculture, and pest control. DL models, particularly convolutional neural networks (CNNs), have proven incredibly effective in precisely detecting and classifying pests in images. This study develops an Innovative Insect Detection and Classification for the Agricultural Sector Using a Gannet Optimization Algorithm with DL (IIDC-GOADL) approach. The main target of the IIDC-GOADL method is to recognize and classify the diverse insect types. The IIDC-GOADL method employs image preprocessing to remove the existing noises. In addition, the densely connected networks (DenseNet) approach is applied for feature extraction. Meanwhile, the GOA can select the optimum hyperparameter for the DenseNet architecture. Moreover, an attention-based bidirectional long short-term memory (ABiLSTM) approach was applied for automated insect discovery and categorization. The experimental outcomes of the IIDC-GOADL method are confirmed under insect datasets, and the results are checked using the dissimilar measures below. An extensive comparison research of the IIDC-GOADL method highlighted an enhanced accuracy outcome of 98.15% and 98.52% over other models under TRP/TSP.
ISSN:2169-3536
2169-3536
DOI:10.1109/ACCESS.2024.3438308