Application of Unsupervised Learning in the Early Detection of Late Blight in Potato Crops Using Image Processing

Abstract

Introduction. Automatic detection can be useful in the search of large crop fields by simply detecting the disease with the symptoms appearing on the leaf.

Objective: This paper presents the application of machine learning techniques aimed at detecting late blight disease using unsupervised learning methods such as K-Means and hierarchical clustering.

Method: The methodology used is composed by the following phases: acquisition of the dataset, image processing, feature extraction, feature selection, implementation of the learning model, performance measurement of the algorithm, finally a 68.24% hit rate was obtained being this the best result of the unsupervised learning algorithms implemented, using 3 clusters for clustering.

Results: According to the results obtained, the performance of the K-Means algorithm can be evaluated, i.e. 202 hits and 116 misses.

Conclusions: Unsupervised learning algorithms are very efficient when processing a large amount of data, in this case a large amount of images without the need for predefined labels, its use to solve local problems such as late blight affectations in potato crops are novel,