Downloads: 5 | Views: 282 | Weekly Hits: ⮙1 | Monthly Hits: ⮙1

Experimental Result Paper | Computer Science & Engineering | India | Volume 11 Issue 12, December 2022 | Popularity: 5.2 / 10

Rate This Article!      

Effective Feature Extraction and Classification Method for Potato Foliar and Tuber Disease Detection using Machine Learning

Megha Rani Raigonda, Sujatha P. Terdal

Abstract: The Indian economy heavily depends on agriculture. Consequently, it is crucial to detect diseases in the agricultural sector. There is a necessity to identify the disease at the initial stage because farmers struggle to produce crops properly because of several plant diseases. The potato is a vegetatively propagated crop. It is a host to many bacterial fungal and viral diseases. A potato crop can be infected by more than 30 plant viruses, a viroid, and phytoplasmas. Viral diseases are of major concern nowadays in potato crops. Once discovered in the field, virus-infected plants result in declassification or even rejection of the seed lots, which results in a monetary loss. The viruses Potato Virus A (PVA), Potato Virus X (PVX), Potato Virus S (PVS), Potato Virus Y (PVY), Potato Virus M (PVM), Potato leaf roll virus (PLRV), and Tomato leaf curl New Delhi virus are known to infect potatoes in India (ToLCNDV), Potato spindle tuber viroid (PSTVD) and Groundnut bud necrosis virus (GBNV). PVM, PVY, PVA, PVX, and PVS occur commonly. Nowadays many computer vision technologies like machine learning, deep Learning are employed in building a prediction model for the effective, rapid, and accurate detection of potato plant disease. In the proposed work the viral disease considered is Potato Leaf Roll Virus (PLRV), Mosaic Virus, Leaf curl, and tuber diseases Potato Tuber Viroid Disease (PSTVD), Potato Virus Y (PVY) Tuber cracking. The foliar image is initially resized to 256* 256, applied contrast enhancement and then filters are applied for denoising removing high frequency, and smoothing the image then the segmentation using Canny Edge Detection is applied to the blurred image to accurately detect the edges of the leaf and then the suitable features are extracted. The disease is classified using classification methods like Support Vector Machine (SVM) and Random Forest. The Random Forest classifier outperforms all other classifiers and produces a classification accuracy of 98.12%.

Keywords: Gray level Cooccurrence matrix, GLCM, Support Vector Machine, SVM, Random Forest, RF, Global Features, GF, Viral Disease

Edition: Volume 11 Issue 12, December 2022

Pages: 717 - 723

Make Sure to Disable the Pop-Up Blocker of Web Browser

Click to Cite this Article

Similar Articles