Development and evaluation of a machine vision system for coconut harvesting and collection
S. Rathinavel*, R. Kavitha, A. Surendrakumar, Ravuri Saiprasanth and M. Suwathiga,
Department of Farm Machinery and Power Engineering, Tamil Nadu Agricultural University, Coimbatore - 641 003, India. Corresponding e-mail: rathinavelesr@gmail.com
DOI: https://doi.org/10.37855/jah.2025.v27i01.25
Key words: Coconut farming, logistics, machine vision, robotics, model
Abstract: The study aimed to develop and analyze a machine vision system for real-time coconut detection to enhance robotic harvesting and collection. Images of healthy and defective coconuts, both green and brown, were captured from Coimbatore and Tirupur districts in Tamil Nadu. These images were processed using a Faster R-CNN model integrated with necessary hardware and software. The system successfully identified the class and grade of coconuts, demonstrating potential applications in robotic harvesting and grading. The model achieved 88% precision and 85% accuracy. Limitations and proposed solutions for the system’s operation are discussed with recommendations for operation-specific measures to improve future robotic developments.
