Agriculture supports living beings and nature in many ways, such as food and habitat. All humans as well as animals depend on the agriculture sector for their basic needs of food and nutrients. Thus, the importance of agriculture and the ways it can be improved further should be taken into account. While discussing agriculture, the topic of related diseases is not far behind. Crop disease is the most common threat that falls upon agricultural harvest. To address the issue at hand, firstly, farmers must figure out the situation on time. Then only the chance of saving the crops exists. We are aware of how the various types of diseases affect the yield badly and in turn affect the quantity. In this paper, we will detect such diseases and identify them. We will use image processing, Machine Learning Algorithms, and sample datasets to get the resulting detection.
In India, where 80% of the population is dependent on agriculture, agribusiness has emerged as a key driver of economic development. Ranchers determine the best harvest based on the soil type, climate, local condition, and economic value. Most often, farmers who cannot repay bank loans they took out for farming purposes commit suicide as a result of production losses. We've seen that the climate is constantly changing today, which harms the crops and pushes farmers into debt and suicide. When various mathematical or statistical techniques are applied to data, these risks can be reduced, and using these techniques, we can inform the farmer about plant diseases affecting his agricultural land.
Due to governmental shortcomings, the agricultural industry started looking for new ways to increase food production. Researchers are working to create innovative, strong, and special technologies that can produce outcomes with high efficacy. A large harvest can be produced using a variety of techniques in the agricultural sector.Based on the data gathered, farmers using the precision agriculture technique can learn how to increase farm yield.
This methodology uses machine learning to classify plant diseases and image processing to detect them. Supervised learning and unsupervised learning are the two main classifications of machine learning. The label values in the supervised learning method are known. Regression and classification are two examples of supervised learning techniques. The label values in the unsupervised learning approach are unknown. Building a model that can be used to categorize healthy and unhealthy harvest leaves and, if the crop has any disease, predict which disease it is, is the primary goal of this research.
II. LITERATURE REVIEW
A review of machine learning classification techniques for plant disease detection was proposed by Shruth et al.  and mainly described the overall plant disease detection system, which can be broken down into five significant stages, including image acquisition, annotated datasets, image processing, feature extraction, and classification.
Image acquisition, image processing, image segmentation, feature extraction, detection, and classification of plant disease are the five steps that make up Sehgal et al" Plant .'s disease classification using soft computing and supervised machine learning" . Data collection or data gathering processes use image acquisition. Some image features can be improved through image processing. The extraction of diseased areas from a plant's leaf uses image segmentation.
The color of the plant's diseased part is extracted using feature extraction.
Different machine learning techniques are used to identify plant diseases using detection and classification techniques.
The article "Image processing techniques for detecting and classifying plant disease: a review" was proposed by Hungilo et al. To identify and classify disease on plant leaves in this proposed system, image processing, and machine learning techniques are used. There are two different approaches to dealing with the classification of plant diseases. The first is "training from the ground up," and the second is "transfer learning." When working with small datasets for deep learning tasks, the transfer learning method is used, and when working with sufficient datasets, the training from the scratch method. This research paper discusses the convolution neural network technique and the conventional technique for image processing.
Leaf diseases can be identified by observing leaf of plants. Image processing is performed on the images of leaves from the dataset. The image processing technique involves five basic steps and a data flow diagram is shown below in Fig 1.
Image Acquisition:This dataset contains high-quality RGB (red, green, and blue) leaf images that include healthy leaf images and diseased leaf images. The first stage of the leaf disease identification and classification system is loading the database for use in the Leaf Dataset
Image Pre-processing:The second stage of the leaf disease detection and classification system is pre-processing. This involves some techniques like image resizing, image smoothing, image enhancement, etc. In this stage, noisy and inconsistent data is removed from the dataset to extract useful information.
Image Segmentation:Images are partitioned into RGB (red, green, blue), HSV (hue, saturation, value), and LAB (luminosity, green/red, blue/yellow) stages. This is the third stage of the leaf illness identification and classification framework.
Feature Extraction in Image:In-leaf disease classification uses features like color, texture, morphology, edges, etc. This approach has tracked down that morphological outcomes give preferable outcomes over different features. It tends to be identifying the infected leaf of the image that is most important.
Detection and Classification of Plant Disease:The fifth stage of the leaf disease recognition and classification system is the classification technique. This stage performs any of the machine learning techniques for classifying the various diseases in leaves. This is the final outcome stage of our proposed system. In this stage, users can identify and characterize leaf diseases.
IV. EXPERIMENTAL RESULT
The experiment made use of a Kaggle data set. The RGB images of healthy and unhealthy leaves in this dataset are arranged into different classes. 20% of the ratio is used for the testing task on this dataset, and 80% of the ratio is used for training. Another index that was later created for prediction purposes is present in test images.
V. FUTURE WORK
In the future we can also suggest Solutions for the detected diseases.
Future work can focus on improving. More datasets may be used in future work.
In the future we can work on raising the accuracy of the results.
The paper mainly focuses on leaf disease detection in various plants. We have used machine learning techniques to train the dataset and also used image processing for the detection and classification of images of both healthy and unhealthy leaves. Thus, various diseases are identified successfully as a result of the proposed system.
 Vishnu S Babu and R. Satheesh Kumar “A Comparative Study on Disease Detection of Plants using Machine Learning Techniques” at 2021 7th International Conference on Advanced Computing and Communication Systems (ICACCS)
 Dhruvi Gosai and Binal Kaka “Plant Disease Detection and Classification Using Machine Learning Algorithm” in 2022 International Conference for Advancement in Technology (ICONAT) Goa, India. Jan 21-22, 2022
 Amrita S.Tulshan and Nataasha Raul “Plant Leaf Disease Detection using Machine Learning” in IEEE Xplore, 10th ICCCNT 2019
 Rajesh Kumar V and Pradeepan K “Identification Of Plant Diseases Using Image Processing And Image Recognition” in IEEE Xplore, IEEE ICSCAN 2021