Abstract

Advances in image processing technology have enabled automation in various aspects of life, including the safety of cosmetic products. This study aims to develop a hazardous ingredient detection system in cosmetic products through label image pattern recognition using deep learning methods. The dataset consists of 500 cosmetic label images containing ingredients, which are then processed using Optical Character Recognition (OCR) to extract text. A deep learning model based on Convolutional Neural Network (CNN) is developed to identify hazardous ingredient keywords in accordance with the regulations of the Indonesian Food and Drug Administration (BPOM) and international organizations such as the FDA and the European Union Cosmetic Regulation. Experimental results show that the proposed model has an accuracy rate of 91.2% in detecting hazardous content compared to conventional text-matching-based methods. This study provides a fast and accurate automated solution to improve consumer safety against cosmetic products containing hazardous ingredients.

Concepts :

Access to Document

Citations by Year