Agriculture is the primary source of livelihood for about more than 50% of the Indian population and rice is one of the major food grains of India. It is observed that rice plant diseases are the major contributors to reduce the production & quality of food. Identification of such diseases may improve the production quality. This paper gives an idea about different methods such as image processing, machine learning & deep learning which are used to detect deadly diseases in rice plants. Much research has been done to automate the rice plant disease detection process using images of the leaf. This manuscript has compared different rice plant disease detection methods and it is found that deep learning methods are more promising than other two methods.
small black linear lesions on leaf blades, leaf tips may turn grey and dry[5]. elongated lesions near the leaf tips and margins, and turns white to yellow and then grey due to fungal attack [5]. dark brown colored and round to oval-shaped lesions on rice leaves[5]. white to gray-green lesions or spots, with dark green borders.
Older lesions on the leaves are elliptical shaped and whitish to gray centers with red to brownish or necrotic borders.
in oval or elliptical are formed irregularly. The enlarged spots become grey combined with white with an outline border in purple brown or blackish brown can be seen[6].
The formation of small seized black lesions found on the sheath of the outer leaf close to the water line, which spread to the sheath of inner leaf resulting in the rotting of tissues[6].
acquisition, pre-processing, image segmentation, feature extraction, and classification. on
be manual capturing of images or some dataset. Most of the researchers have captured images using a camera with high resolution in paddy fields and then resized the images into some definite number of pixels[7]–[9]. On the other hand, the dataset named UC Irvine Machine Learning Repository [1] and imageNet [10] is also used for image acquisition.
that is more meaningful and less complex to analyze. Here, an image can be divided into some regions based on the desired feature[12]. Several segmentation techniques are the Otsu segmentation method, K-Means Clustering, region segmentation, contours, etc.
Classification is an important module in plant disease detection systems. It is defined as a process of categorizing plant leaf images based on identified diseases. There are two main classification techniques namely supervised and unsupervised. In Supervised classification, we have pre-trained data that helps to predict outcomes for some unforeseen data. The trained classifier is used to group different pictures. The Unsupervised order utilizes the properties of the pixels to bunch them, these gatherings known as a group, and process called clustering[15]. Some supervised classification algorithms like Logistic Regression, K-Nearest Neighbor, Decision Tree, Naive Bayes were applied for classification[16]. Also, the artificial neural network is one of the emerging methods of classification.
disease identification of paddy crops are presented along with their performance measure (Table II). Most researchers have identified four major diseases of rice plants: blast, bacterial blight, spot, and leaf smut. Different segmentation and classification techniques are used for detecting these diseases. K-means clustering and Otsu’s method are giving significant results for segmentation. Further, it is observed that deep neural network and decision tree classifiers are giving the highest accuracy of >=97% for identifying rice plant diseases. Several researches have been done to find the optimum solution to identify most prominent diseases in the rice plant[17].
optimum solution for this problem. Different techniques are applied on diseased rice images so that further research can be made in this area to improve the overall performance of the rice disease detection system. This paper reviewed and summarized techniques of image processing, machine learning and deep learning that have been used in disease identification.
It is found that extraction of the affected region from the leaf image is the utmost important step, for which we have studied different segmentation techniques. A comparison between different methodologies for rice disease detection has been made and it can be concluded that deep neural network and decision tree classifiers are giving highest accuracy of >=97% for identifying diseases in rice crop. Still there is a need to work to identify more rice plant diseases other than the four major diseases which are discussed in this paper. 6. References