<![CDATA[This study aims to develop a web-based application that integrates You Only Look Once version 10 (YOLOv10) and Optical Character Recognition (OCR) to analyze and visualize nutritional value information tables on packaged products. The system extracts text using OCR and presents a visualization of nutritional content in easy-to-understand terms, including comparisons to recommended daily intake as well as foods with equivalent nutritional content. The YOLOv10-M model for nutrient table detection achieved high accuracy with a mean Average Precision value at the 50% threshold (mAP50) of 0.995, while the YOLOv10-S model for nutrient value detection showed moderate performance with a mAP50 of 0.731. EasyOCR is used in the text extraction process with the best results in bright background conditions (90.09%) and lowest in uneven surfaces (26.48%). The results of the system test showed a level of functionality of 100%, with an average System Usability Scale (SUS) score of 73.3 and a Net Promoter Score (NPS) of 31.82%, both of which are included in the "good" category. These findings show that the developed system can help consumers understand product nutrition information more easily, as well as potentially increase public awareness of healthy consumption patterns.]]>
Copyrights © 2025
