A neural-network based algorithm oriented to identifying the damage degree caused by the Meloidogyne Incognita Nematode in Digital Images of Vegetable Roots

This work proposes an algorithm oriented towards detecting the amount of damage or infection caused by the Meloidogyne Incognita nematode, through the extraction of physical features in digital images of vegetable roots. The aim is to reduce the subjectivity in sample analysis by visual inspection made by specialized personnel, and to reduce the sample analysis time. The algorithm consists of a thresholding step, a filtering step, labeling and physical feature extraction. Next, the obtained data feeds a neural network, which determines the infection level through the Zeck scale. For the validation process, samples were selected whenever 2 specialists gave the same infection score. Results showed a 98.62% specificity level and a 93.75% sensitivity level.