Extreme learning machine with feature extraction using GLCM for phosphorus deficiency identification of cocoa plants

This study aims to analyze the implementation of the Extreme Learning Machine (ELM) Algorithm with Gray Level Co-Occurrence Matrix (GLCM) as an Image Feature Extraction method in identifying phosphorus deficiency in cocoa plants based on leaf characteristics. Characteristic images of cocoa leaves were placed under normal conditions and phosphorus deficiency, each with 250 datasets. The feature extraction process by GLCM was analyzed using the ELM parameter approach in the form of Network Node_Hidden variations and several Activation Functions. The method of this case study was conducted with data collection, algorithm development to validation, and measurement using ROC. It was found that the best accuracy when testing the dataset was 95.14% on the node_hidden 50 networks using the Multiquadric Activation Function. These results indicate that the feature extraction model with GLCM using Contrast, Correlation, Angular Second Moment, and Inverse Difference Momentum properties can be maximized on Multiquadric Activation Function.