Real-Time System for Detecting Harmful Ingredients in Underarm Products Using Optical Character Recognition and Machine Learning Techniques

Personal hygiene products, such as underarm products, deodorants, and antiperspirants, are essential for daily use. However, many people are concerned about the potential health effects of the unfamiliar ingredients listed on these products. Existing applications that help consumers check product safety often require account creation, contain advertisements, or provide unhelpful results. As a result, users aim to search manually for each ingredient on the label, which can be time-consuming. Therefore, this study aims to develop a real-time system that can quickly detect and accurately identify harmful chemical substances; Propylene glycol, and Aluminum in underarm products. The system applied Optical Character Recognition (OCR) to extract ingredient names and a Support Vector Machine (SVM) for classification. The system’s ability to detect with 93.4% accuracy in classifying 120 test samples using the SVM classifier. For future work, the system’s dataset can be expanded to include more chemical compounds that may have harmful effects. Furthermore, various machine learning methods and OCR tools could be evaluated to enhance performance, classification accuracy, and text detection in subsequent research. The proposed system provides a potential solution to offer consumers a quick and reliable method to determine the safety of underarm products, and it can help consumers make informed decisions about the safety of the products they use every day.