International Journal of Advances in Engineering & Scientific Research
Print ISSN : 2349 –4824
Online ISSN : 2349 –3607
Frequency : Continuous
Current Issue : Volume 10 , Issue 1
2023
Mr.R. Madanachitran ,Mukeshkumar C, Santhru S, Prashna M
Mr.R. Madanachitran ,Mukeshkumar C, Santhru S, Prashna M, Department of Computer Science and Engineering, Kongunadu College of Engineering and Technology, Tamilnadu, India
DOI : https://doi.org/10.5281/zenodo.14924841 Page No : 17-27
Published Online : 2023-06-30
Download Full Article : PDF Check for Updates
Pest identification plays a crucial role in agricultural pest management, influencing pesticide selection and crop protection strategies. This study introduces Deep Pest Net, a deep learning-based model designed for efficient pest classification and pesticide recommendation. The proposed methodology consists of four key steps: data augmentation, image resizing, dataset partitioning, and model training/testing. To overcome data scarcity, augmentation techniques such as rotation, scaling, and translation were applied, enhancing model generalization. The DeepPestNet architecture comprises eleven learnable layers, including eight convolutional layers and three fully connected layers, optimized with Leaky ReLU and ReLU activation functions. The model was trained and validated using two benchmark datasets: the Deng et al. (2018) dataset (ten pest classes) and the Kaggle "Pest Dataset" (nine pest classes). The experimental results demonstrated 99.96% accuracy, 99.66% precision, 100% recall, and 99.82% F1-score, outperforming existing models such as GoogLeNet and SqueezeNet. Additionally, DeepPestNet was validated on medical imaging datasets, achieving over 96% accuracy, highlighting its cross-domain adaptability. The study concludes that DeepPestNet is a lightweight, efficient, and accurate solution for real-time pest identification, addressing key challenges like overfitting and data scarcity while offering potential applications in agricultural and medical domains.
Keywords: Deep Learning, Pest Classification, Pest Detection, Agricultural Systems, DeepPestNet, CNN, Data Augmentation, Image Processing, Machine Learning.
