A method for detecting and segmenting infected part of cacao pods

• Main Authors: Tan, Daniel Stanley, Leong, Robert Neil, Laguna, Ann Franchesca, Ngo, Courtney Ann M., Lao, Angelyn, Amalin, Divina, Alvindia, Dionisia
• Format: text
• Published: Animo Repository 2016
• Subjects: Image segmentation; Machine learning; Cacao—Diseases and pests; Computer Sciences
• Online Access: https://animorepository.dlsu.edu.ph/faculty_research/8417
• Institution: De La Salle University

Farmers and agricultural technicians regularly monitor the well-being of their crops. But at present they rely on visual inspection to assess the degree of infestation of their crops, resulting to several errors and inconsistencies due to the subjective nature of the assessment procedure. To improve the inspection procedure, this research shows a method for detecting and segmenting the infected parts of the cacao pods based on K-means algorithm supplemented by a Support Vector Machine (SVM) using image colors in L*a*b* color space as features. The highest attained accuracy was 89.2% using four clusters. Results of this research provides promise in the implementation of the proposed framework in developing a more accurate assessment of infestation level; thus, potentially improving decision support for managing cacao diseases.