Guardat en:
| Autors principals: | , , , , |
|---|---|
| Format: | Recurso digital |
| Idioma: | anglès |
| Publicat: |
Zenodo
2026
|
| Matèries: | |
| Accés en línia: | https://doi.org/10.5281/zenodo.20032212 |
| Etiquetes: |
Afegir etiqueta
Sense etiquetes, Sigues el primer a etiquetar aquest registre!
|
Taula de continguts:
- <p>This project presents an AI-based system for analyzing the safety of cosmetic products using ingredient information extracted from product labels. The system leverages Optical Character Recognition (OCR) to extract ingredient text from uploaded images and applies machine learning techniques to evaluate toxicity risks.</p> <p>The extracted ingredients are processed using fuzzy matching to correct OCR errors and mapped to a structured toxicological dataset containing scientific attributes such as carcinogenicity (IARC, NTP, ECHA), reproductive toxicity, endocrine disruption, sensitization potential, allergen classification, and regulatory restrictions.</p> <p>A trained machine learning model predicts toxicity scores for each ingredient, and a product-level safety score is computed based on the highest-risk ingredient, following established toxicological principles. The system provides explainable outputs and visual representations including bar charts, pie charts, and ingredient-level analysis through an interactive React-based dashboard.</p> <p>The methodology is inspired by the Think Dirty® rating system and is grounded in the principle:<br>Risk = Hazard × Exposure.</p> <p>The system aims to improve consumer awareness by providing transparent, scientific, and real-time cosmetic safety insights.</p>