FoodX: An AI-Powered System for Packaged Food Ingredient Analysis and Personalized Health Recommendations

Abstract

This research presents a technology-driven solution aimed at simplifying the understanding of nutritional information in packaged food products. With the increasing consumption of processed food and rising health concerns, there is a growing need for tools that enable informed dietary decisions. The proposed system integrates optical character recognition (OCR), machine learning (ML), and natural language processing (NLP) to extract, analyze, and personalize nutritional data from product labels. By incorporating user-specific health profiles and real-time image processing, the system offers personalized health insights, alternative recommendations, and educational content to encourage healthier eating habits. Designed with a focus on accessibility, this solution aims to bridge the gap between food packaging and public understanding, promoting proactive health management and consumer empowerment through intelligent food analysis.

Introduction

Overview:
Food plays a vital role in human health, yet the increasing consumption of unhealthy packaged foods is leading to chronic conditions like diabetes and obesity. Many packaged items contain hidden harmful ingredients due to misleading labels. FoodX is an AI-driven solution aimed at simplifying nutritional analysis and promoting healthier food choices, particularly for individuals with specific health conditions.

A. Problem Statement

FoodX uses AI, OCR (via Google Vision API), and an Artificial Neural Network (ANN) to extract and evaluate nutritional information from packaged food images. It identifies harmful ingredients and classifies food safety levels using a React Native frontend, Node.js backend, and MongoDB Atlas database.

B. Motivation

Aggressive marketing misleads consumers, especially children, into consuming unhealthy products. Despite awareness campaigns, tools enabling informed dietary choices are lacking. FoodX bridges this gap by shifting public behavior from reactive to proactive health management through real-time analysis and educational visualizations.

C. Objectives

1. Automate extraction and verification of food label claims.

2. Use AI to identify harmful ingredients.

3. Predict long-term health risks (e.g., obesity, heart disease).

4. Provide personalized health recommendations based on user history.

D. Scope

FoodX combines food safety, health awareness, and technology to help users make informed food choices. It analyzes nutritional content using OCR and ML, predicts health risks, and suggests safer alternatives. NLP is employed to generate educational content, enhancing digital food literacy.

II. Literature Review

Studies show rising concern over packaged food safety in India, particularly among low-income groups who lack label awareness. AI and ML, including ANN and OCR technologies, can identify hidden harmful components and support health prediction. Personalized Health Recommendation Systems (PHRS) enhance user-specific dietary advice. NLP further supports personalization by extracting and presenting relevant information clearly.

III. Methodology

A. Image-Based Nutrition Analysis

1. Image Capture & Encoding: Users upload images of food labels, encoded in Base64 for secure transmission.

2. OCR Processing: Google Vision API extracts text after preprocessing (noise removal, skew correction).

3. Text Analysis: Text is cleaned, parsed using regex, and nutritional values are normalized.

4. ANN Classification:

   • Nutritional features are fed into an ANN model.

   • Food is classified into Safe, OK, Harmful, or Very Harmful.

5. User Output:

   • Cleaned nutritional data, safety labels, interpretation messages, and health references are provided.

B. NLP-Driven Awareness Generation

1. Data Aggregation: Collects user scan data and identifies trends (e.g., frequent high-sodium foods).

2. Topic Generation: NLP models use trends to write informative, personalized articles on nutrition (e.g., health effects of sugar).

3. Content Delivery: Articles aim to improve user literacy and support long-term health behavior change.

Conclusion

This paper presents a comprehensive approach to packaged food analysis and personalized health recommendation by integrating OCR, machine learning, and natural language processing techniques. By addressing key limitations in existing systems—such as lack of personalization, poor label recognition accuracy, and limited user awareness tools—our proposed model aims to empower consumers with real-time, health-conscious food choices. The inclusion of features like ingredient classification, medical history-based filtering, and NLP-generated educational content ensures that the system is not only technically robust but also user-centric and impactful for public health awareness.

References

[1] A review on food recognition technology for health applications Dario Allegra, Sebastiano Battiato, Alessandro Ortis, Salvatore Urso, Riccardo Polosa Department of Mathematics and Computer Science; Center of Excellence for the Acceleration of Harm Reduction (CoEHAR), University of Catania, Catania, Italy (2020) [2] \'JagoGrahakJago\': A cross-sectional study to assess awareness about food adulteration in an urban slum Medical Cadet Shailesh Ishwar, Lt Col Puja Dudeja, Maj Pooja Shankar, Surg Lt Cdr Santosh Swain, Air Cmde Sandip Mukherji (2016) [3] Aiding food security and sustainability efforts through graph neural network?based consume food ingredient detection and substitution Jack Foster & Alexandra Brintrup. (2023) [4] Extraction of Text from Images Using Deep Learning, M Sinthujaa, Chirag Ganesh Padubidri, Gaddam Sai Jayachandrac, MudduluruCharanTejad, Golthi Sai Pavan Kumar. (2023) [5] Automated Data Digitization System for Vehicle Registration Certificates Using Google Cloud Vision API KaranratThammarak, YaowaratSirisathitkul, PrateepKongkla, SarunIntakosum (2022) [6] Artificial Neural Network: A brief study Mayuri Thorat, Shraddha V. Pandit, SupriyaBalote (2023) [7] Artificial Neural Networks for Predicting Food Antiradical Potential Victor Gorbachev, Marina Nikitina, Daria Velina, SherzodkhonMutallibzoda, Vladimir Nosov,GalinaKorneva, Anna Terekhova, Elena Artemova, Bella Khashir, Igor Sokolov, Svetlana Dimitrieva and Igor Nikitin (2022) [8] Role of different nutrient profiling models in identifying targeted foods for front-of-package food labelling in Brazil Ana Clara Duran, Camila Zancheta Ricardo, Laís Amaral Mais Ana Paula Bortoletto Martins (2020) [9] Generating and Analyzing Chatbot Responses using Natural Language ProcessingMoneerhAleedy, HadilShaiba (2019) [10] Nutrition and health claims on healthy and less-healthy packaged food products in New Zealand Haya H. Al-Ani, Anandita Devil, Helen Eyles, Boyd Swinburn1 and Stefanie Vandevijvere (2016) [11] Road map to enter Indian nutraceuticals and health supplements business, Pradip Chakraborty Food Safety and Standards Authority of India, New Delhi, India. (2019) [12] Innovative approaches to food quality control: AI and machine learning for predictive analysisTemiladeAbass, Esther OleiyeItua, TabatBature and Michael AlurameEruaga (2024) [13] OCR with Tesseract, Amazon Textract, and Google Document AI: a benchmarking experiment Thomas Hegghammer (2021) [14] Nutrition labels on pre-packaged foods: a systematic review Sarah Campos, Juliana Doxey and David Hammond (2011) [15] Global trends in added sugars and non-nutritive sweetener use in the packaged food supply: drivers and implications for public health Cherie Russell1, Phillip Baker, Carley Grimes, Rebecca Lindberg and Mark A Lawrence (2022) [16] ANN Based Multiclass Classification of P2P Botnet Chirag Joshi, Ranjeet K Ranjan2 and Vishal Bharti (2022) [17] Introduction to artificial neural networks Enzo Grossi (2008) [18] Extracting Information from Vehicle Registration Plate using OCR Tesseract Agung YuwonoSugiyonoa, KendrickoAdrioa, Kevin Tanuwijayaa, Kristien Margi. (2023) [19] Machine learning prediction of the degree of food processing, Giulia Menichetti, Babak Ravandi, DariushMozaffarian, Albert-László Barabási. (2023) [20] Multi-food detection using a modified swin-transformer with recursive feature pyramid network Chao-Yang Lee1, Abida Khanum, PinnintiPraneeth Kumar. (2023) [21] Awareness And Knowledge Of Nutrition Labels On Pre-Packaged Food Items Among Urban Adolescents Of Delhi, IndiaPreeti Kamboj University of Delhi, Neena Bhatia Lady Irwin College. (February 2022) [22] Application of machine learning to the monitoring and prediction of food safety: A review Xinxin Wang, YamineBouzembrak, AGJM Oude Lansink, H. J. van der, Fels-Klerx. (2022) [23] Design an Archetype to Predict the impact of diet and lifestyle interventions in autoimmune diseases using Deep Learning and Artificial Intelligence, Abeda Begum, Bhagwant University, Rajeev Kumar, Chandigarh University. (2021) [24] Artificial Intelligence in Nutrients Science Research: A ReviewJaros?awSak and Magdalena Suchodolska. (2021) [25] A review on food recognition technology for health applications Dario Allegra, Sebastiano Battiato, Alessandro Ortis, Salvatore Urso, Riccardo Polosa. (2020)

Copyright

Copyright © 2025 Dr. Deepali Ujalambkar , Kunal Kamat, Gayatri Kulkarni, Prajwal Khedkar, Kaustubh Korpe. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.