Dehusked Coconut Vision-Based Counting on a Manufacturing Plant Utilizing the YOLOv8, ByteTrack, and Roboflow Algorithms

Many manufacturing industries in third-world countries are still in need of process systems improvement in order to increase productivity. Some manufacturing plants are encountering inconveniences in reconstructing establishments to apply advance technologies, thereby forcing engineers to provide cheaper and convenient solutions that are slightly incompetent rather than the more effective ones. In line with this, the present study aimed to make cameras with computer vision (CV) applications as a replacement for commonly used expensive sensors in coconut quantity monitoring to improve competency and accuracy. You Only Look Once version 8, ByteTrack, and Roboflow Supervision was used on developing a dehusked-coconut vision-based detecting, tracking, and counting model. The study developed a 97.9% accurate model, enough to become basis for a production analysis on manufacturing process. Low-resolution images and complex backgrounds cause inaccuracy when the functionality of the model was evaluated. This model will assist coconut manufacturing plants in meeting market demands.