DETERMINATION OF LEAF RUST (Puccinia Allii Rudolphi) SEVERITY AND ITS MANAGEMENT ON GARLIC (Allium Sativum L.) USING IMAGE PROCESSING TECHNIQUES

Abstract

Traditionally, fungicides are applied manually without objectively quantifying the severity of garlic rust disease. In this research, digital image processing techniques were employed to develop and test an algorithm that could measure the severity of leaf rust on garlic and manage the disease at various levels. Data on leaf rust were collected from seven groups of garlic plants, with samples taken from the top, middle, and lower leaves. Each group consisted of three pots, each containing two plants after thinning. The control group received no fungicide (Tilt or Propiconazole), while group 4 was treated with the recommended 60 mL of Tilt fungicide. Groups 1-3 received 10%, 5%, and 0% less than the recommended volume (6 mL, 3 mL, and 0 mL, respectively). Groups 5-7 received 5%, 10%, and 100% more than the recommended volume (3 mL, 6 mL, and 60 mL, respectively). The fungicide was sprayed on garlic leaves four times to observe the treatment progression and optimize the fungicide volume. A total of 504 garlic leaf images were captured across four rounds: before spraying fungicide, after the first spray, after the second spray, and after the third spray. The total leaf area and the diseased areas were extracted from these images. The relative error between experts' assessments and the imaging algorithm was found to be 4.2%. The algorithm developed for estimating disease severity using image processing technology demonstrated an accuracy of 95.80%. The results indicate the potential of this technology for measuring garlic rust severity and optimizing fungicide application.