DEEP LEARNING APPROACH FOR CROP DISEASE DETECTION AND CLASSIFICATION WITH PESTICIDE SUGGESTION

Abstract

A nation's innovation depends on its agriculture. All nations are built on agriculture, which provides food and raw materials. Food from agriculture is vital to humans. Detecting plant diseases has become a priority. With powerful computing systems and large datasets, deep learning in crop disease detection became popular in the early 21st century. Farmers used manual observation and generational knowledge to identify crop diseases in the traditional system. Agricultural experts would inspect crops, diagnose diseases by symptoms, and recommend treatments. This method was useful, but it was time-consuming, dependent on the observer, and sometimes misdiagnosed. As the global population grows and food demand rises, crop disease detection requires advanced methods like deep learning. Preventing yield losses requires timely and accurate crop disease identification. Farmers can quickly control disease spread by automating detection, increasing agricultural productivity. By reducing chemical use, precise pesticide recommendations reduce farming's environmental impact. Convolutional neural networks (CNNs) are excellent image recognition algorithms, making them ideal for identifying patterns in diseased crop images. Deep learning for crop disease detection and classification has transformed crop management. Farmers can now detect crop diseases more accurately and efficiently using advanced technologies. This allows timely intervention and suggests disease-controlling measures like pesticide use, which affects food security.

Specification

Description:Agriculture is one of the most important sources for human sustenance on Earth. Not only does it provide the much necessary food for human existence and consumption but also plays a major vital role in the economy of the country. But Plant diseases have turned into a dilemma as it can cause significant reduction in both quality and quantity of agricultural products. Nowadays farmers are facing many crucial problems for getting better yield cause of rapid change in climate and unexpected level of insects, in order to get better yield, need to reduce the level of pest insect. Several millions of dollars are spent worldwide for the safety of crops, agricultural produce and good, healthy yield. It is a matter of concern to safeguard crops from Bio-aggressors such as pests and insects, which otherwise lead to widespread damage and loss of crops. In a country such as India, approximately 18% of crop yield is lost due to pest attacks every year which is valued around 90,000 million rupees. Conventionally, manual pest monitoring techniques, sticky traps, black light traps are being utilized for pest monitoring and detection in farms.
Manual pest monitoring techniques are time consuming and subjective to the availability of a human expert to detect the same. Disease is caused by pathogen which is any agent causing disease. In most of the cases pests or diseases are seen on the leaves or stems of the plant. Therefore, identification of plants, leaves, stems and finding out the pest or diseases, percentage of the pest or disease incidence, symptoms of the pest or disease attack, plays a key role in successful cultivation of crops. In general, there are two types of factors which can bring death and destruction to plants; living(biotic) and nonliving (abiotic) agents. Living agent's including insects, bacteria, fungi and viruses. Nonliving agents include extremes of temperature, excess moisture, poor light, insufficient nutrients, and poor soil pH and air pollutants.
In recent years, deep learning has made breakthroughs in the field of digital image processing, far superior to traditional methods. How to use deep learning technology to study plant diseases and pests' identification has become a research issue of great.
Crop disease datasets are pre-processed and uploaded to Residual Network-CNN ((ResNet-CNN) for feature extraction. On the other hand, leaf images are also pre-processed and uploaded to ResNet CNN for testing. The leaf images and the crop disease datasets are compared to the trained features which are already trained with the plant diseases. The extracted features have some loss computation and accuracy. The comparison graph could predict the classes of the plant disease.
Deep neural network is gradually applied to the identification of crop diseases and insect pests. Deep neural network is designed by imitating the structure of biological neural
Network, an artificial neural network to imitate the brain, using learnable parameters to replace the links between neurons. Convolutional neural network is one of the most widely used deep neural network structures, which is a branch of feed forward neural network. The success of AlexNet network model also confirms the importance of convolutional neural network model. Since then, convolutional neural networks have developed vigorously and have been widely used in financial supervision, text and speech recognition, smart home, medical diagnosis, and other fields.
Convolutional neural networks are generally composed of three parts. Convolution layer for feature extraction. The convergence layer, also known as the pooling layer, is mainly used for feature selection. The number of parameters is reduced by reducing the number of features. The full connection layer carries out the summary and output of the characteristics. A convolution layer is consisting of a convolution process and a nonlinear activation function ReLU. A typical architecture of CNN model for crop disease recognition is shown in Fig. 1.
The leftmost image is the input layer, which the computer understands as the input of several matrices. Next is the convolution layer, the activation function of which uses ReLU. The pooling layer has no activation function. The combination of convolution and pooling layers can be constructed many times. The combination of convolution layer and convolution layer or convolution layer and pool layer can be very flexibly, which is not limited when constructing the model. But the most common CNN is a combination of several convolution layers and pooling layers. Finally, there is a full connection layer, which acts as a classifier and maps the learned feature representation to the sample label space.
, C , C , Claims:
1. We claim the deep learning model claims to achieve high accuracy and precision in detecting and classifying a wide range of crop diseases from images, outperforming traditional diagnostic methods.
2. We claim the system is capable of providing real-time analysis, enabling immediate disease detection and timely intervention to minimize crop losses.
3. We claim by integrating a comprehensive database, the approach claims to offer automated pesticide suggestions tailored to the specific disease, crop type, and environmental conditions, enhancing the effectiveness of treatments.
4. We claim the system is designed to be intuitive and easy to use, allowing farmers to upload images and receive instant feedback without requiring extensive technical knowledge.
5. We claim the model claims to be adaptable for use across different types of crops and geographical locations, ensuring its applicability in diverse agricultural practices.

Documents