A Hybrid Meta Model for Detecting Cotton Disease Employing an IoT-based Platform and an Ensemble Learning Methodology

Bhushan V. Patil^1*and Pravin S. Patil²

¹Electronics and Telecommunication Engineering, R. C. Patel Institute of Technology, Shirpur, India.

²Electronics and Telecommunication Engineering, Shri Shivaji Vidya Prasarak Sanstha's Bapusaheb Shivajirao Deore College of Engineering, Dhule, India.

Corresponding Author E-mail: patilbhushan007@gmail.com

DOI : http://dx.doi.org/10.12944/CARJ.12.2.13

Abstract:

Every nation's development depends on its agriculture. "Cash crops" are important crops like cotton and others. Most pathogens that cause substantial crop damage also affect cotton. Many illnesses have an impact on yield through the leaf. Recognizing illnesses early causes additional damage to crops. Many diseases can harm cotton, including as powdery mildew, leaf curl, bacterial blight, leaf spot, target spot, and nutrient deficiencies. Accurate disease detection is necessary before the proper course of treatment can be taken. Accurate plant disease diagnosis depends on deep learning. With accuracy, the suggested model, which is based on meta-Deep Learning, can identify different cotton leaf diseases.  When utilizing an Internet of Things (IoT)-based sensor technology to identify cotton plant diseases, prior information about soil moisture levels, relative humidity, temperature, leaf wetness, and rainfall is crucial. For this study in real time, we collected sensor-based information and 1956 pictures of cotton leaves that were cultivated in the field, showing both sick and healthy leaves. The data augmentation method increased the size of the dataset. The dataset was trained with Custom CNN to get good accuracy for cotton diseases prediction, And Classification is carried out using a stacking ensemble model, which combines, ResNet50, VGG16, and InceptionV3 models for more accurate Disease prediction.

Keywords:

CNN; Deep Learning; Ensemble learning; Hybrid classifier; IoT; Meta-model

Download this article as: 

[ HTML Full Text]