<![CDATA[Machine vision is a technology commonly used in modern industry for image-based autonomous analysis and inspection. Machine vision helps the process of product analysis and inspection in the industry faster than manual analysis and inspection. This study applies machine vision to an image processing-based product number identification system in the manufacturing industry packing process using the YOLOv4 algorithm and evaluation of the confusion matrix system. The results of the identification are stored in a database and displayed on the website to facilitate the monitoring process. This system has carried out several tests, especially testing the main function of the system, namely product calculations, carried out 10x experiments. Then, testing variations in light intensity with a range of 20 – 225 lux and variations in height with a range of 48 – 68 cm with 10 trials each. From the tests that have been carried out, an evaluation of the confusion matrix is applied and produces an accuracy and precision of 100% and an error of 0%. The average computing speed of this system is 6.95 FPS with the help of CUDA.]]>
Copyrights © 2024
