Crop Disease Identification Using Deep Learning Techniques
Crop Disease Identification Using Deep Learning Techniques
Crop Disease Identification Using Deep Learning Techniques
Abstract:- Detection and control of plant diseases is investigating sustainable, environmentally friendly
critical to maintaining global food security. Recent alternatives, such as plant growth-promoting rhizobacteria
advances in deep learning and computer vision have (PGPR), to increase agricultural productivity and get the
revolutionized precision agriculture, especially in most out of farming while minimizing the ecological
automatic detection of crop diseases. This research aims footprint. In short, the emergence of permaculture, driven by
to further advance this new trend using deep learning learning technology and environmentally friendly methods,
techniques. It focuses specifically on the use of was a major turning point in the history of agriculture. The
convolutional neural networks (CNN), specifically the integration of smart agriculture, digital connectivity and
VGG19 architecture, for the accurate and efficient sustainability not only solves today's agricultural problems,
detection of agricultural diseases. The study utilized a but also opens the door to more efficient operations,
large database containing numerous photographs of efficient and environmentally friendly agriculture.
healthy and diseased plants. Adding this information
increases the power and capabilities of the model. The II. LITERATURE REVIEW
VGG19 architecture is based on algorithms that use
transfer learning techniques to extract complex Plant disease diagnosis based on deep learning models.
information from images. This article provides an in-depth look at deep learning
models for plant disease identification. It shows the high
Keywords:- Agriculture, Detection, Rice Disease, Iot accuracy achieved by this method in detecting pests and
Architecture System. diseases affecting plants. [1]
Fig 1 Sample images from Plant Village dataset for different types of leaf diseases.
Training Model and Validation issues. Adopting these methods can ensure the stability and
Selection of training model on given data using reliability of crop diseases.
appropriate optimizer and error. Data sets are often divided
into training sets, validation sets, and test sets to evaluate Technology Introduction
model performance. During training, measures such as early Shallow matching model using random forest and
stopping, and sample checking are used to reduce the drop XGBOOOST.
rate and ensure the best sample weight for delivery is
maintained. Combining shallow VGG model and random forest
(RF) with XGBoost algorithm can be successful in disease
Hyperparameter Tuning diagnosis. This scheme uses VGG topology, RF and
Fine-tuning model hyperparameters such as learning XGBoost algorithms to improve detection accuracy and can
rate, batch size, and activation function greatly affects model provide efficient and cost-effective solutions.
performance. Use exact or stochastic search strategies to
improve these parameters. Effective adaptive methods for crop disease detection
Deep Learning Models With Transfer Learning For Rice The suggested strategies emphasise the employment of
Leaf Diseases multiple models, algorithms, and methodologies, stressing
Deep learning models like CNN-VGG19, applying the significance of deep learning, transfer learning, and
transfer learning, allow exact diagnosis and classification of upgrades in dataset quality for accurate and fast crop disease
rice leaf diseases. These models exploit transfer learning's diagnosis. These technologies provide prospective options
ability to adapt pre-trained models to rice leaf disease for upgrading agricultural practices by simplifying early
detection, assuring accurate disease categorization. disease control and boosting crop output.
Fig 4 Fine-tuned transfer learning for the VGG19 model for rice leaf disease identification.
Vgg-19 For Plant Disease Classification Shallow Convolutional Models For Plant Disease
Research includes implementing VGG-19, a Identification
Convolutional Neural Network (CNN) model, using transfer Experiments presented shallow VGG models linked
learning for predicting disease classes in plants. The with XGBoost and Random Forest (RF) for disease
research aims to accomplish early diagnosis and diagnosis in plants. Utilizing VGG19, the models attempted
categorization of plant diseases using deep learning to reliably detect different plant diseases.
methods.
Vgg-19 Model With Transfer Learning For Tomato Leaf V. RESULTS AND DISCUSSIONS
Diseases
The experiment deployed a VGG-19 model with The adoption of the VGG19 model has shown
transfer learning, reaching an accuracy of 95.48% in excellent results in reliably diagnosing crop diseases,
diagnosing tomato leaf diseases. The technique exceeded obtaining a noteworthy accuracy of 92.4%. This high
typical machine learning networks in illness detection. accuracy is a testimony to the durability and effectiveness of
deep learning approaches in plant disease detection,
Hybrid Model For Plant Disease Detection demonstrating their promise for real-world use in
Experimentation featured a hybrid model that agricultural contexts.
efficiently exploited a minimal number of training
parameters to identify Bacterial Spot disease in peach plants. The importance of this breakthrough rests in its
The suggested model attained an outstanding accuracy of practical ramifications for the agricultural business.
99.35%. Accurate and prompt diagnosis of agricultural diseases is
critical for farmers to prevent possible losses and promote
Improved Crop Disease Identification With Vgg19 And optimum crop health. The VGG19 model, utilizing its deep
Alexnet learning capabilities, provides a reliable and efficient tool
Experiments contrasted VGG19 with AlexNet models, for this goal.
displaying VGG19's accuracy of 92.4% for the training set,
89.9% for the verification set, and 87.8% for the test set in
crop disease detection.
Fig 5 Performance analysis of Inception VGG19 model using Plant Village dataset.
(a) Model recognition accuracy; (b) train and test loss.
One of the key benefits of using deep learning models models can help farmers use more targeted and effective
like VGG19 is their ability to process and analyze large treatments. This can support sustainable agriculture by
amounts of data quickly and reliably. This rapid operation reducing reliance on extensive treatments and optimizing the
leads to timely diagnosis and classification of crop diseases, use of resources such as pesticides.
supporting farmers in timely selection of disease control.
In addition, the effectiveness of the VGG19 model in
In addition, the scalability of deep learning models diagnosing diseases represents a step forward in the use of
makes it possible to adapt to different types of crops and technology in solving agricultural problems. As technology
diseases. This performance is important in agriculture continues to advance, integrating deep learning models into
because many crops will suffer from specific diseases, each agriculture can pave the way for innovations in crop
requiring diagnostic and treatment technology. The management and production.
development of the VGG19 model in keeping many records
and classifying diseases very well contributes to its benefits However, although the VGG19 model achieves high
in agriculture. accuracy, there may be problems in its actual application.
Factors such as dataset diversity, model generalization to
The use of deep learning models such as VGG19 is different environmental conditions, and the need to
also compatible with the spread of precision agriculture. By continually update models to include new diseases may
providing accurate and targeted disease diagnosis, these
represent practical challenges for continued research and improve disease classification accuracy by combining the
development. output of multiple models 3.