FoodTrust AI: A Personalized Food-Label Analysis System with Claim Verification, Package Compliance, and Consumer Reporting Assistance

Abstract

Packaged-foodlabelscarrydetailsaboutingredients, nutrition, allergens, and product claims, but many consumers find this information hard to interpret. This paper presents FoodTrust AI, an Android application that makes food-label informationeasiertounderstandandhelpsuserscheckselected packagedetails.TheapplicationusesCameraXtocapturelabel images and Google ML Kit Text Recognition to extract text. Java-based rules contribute to the health score, while the Gemini API explains the extracted information and provides profile-basedwarningsandrecommendations.Userscancreate their own and family-member profiles, which are managed through Firebase Realtime Database. The application also includes Claim Verification, Package Compliance Checking, and Consumer Reporting Assistance. When a possible issue is found, the app prepares a report document and asks for the user’s confirmation before opening the relevant government reporting website. The main contribution is combining personalizedfoodanalysisandconsumer-complianceassistance in one Android application. Initial tests were carried out on selected real packaged-food products; numerical evaluation of OCRaccuracy, scoreconsistency, recommendationquality,and compliance detection has not yet been completed.

Introduction

Food packages contain important information about ingredients, nutritional values, allergens, and product claims. However, technical terminology, small print, and unfamiliar ingredients often make it difficult for consumers to understand food labels and make informed dietary decisions. FoodTrust AI is an Android-based mobile application developed to address this problem by combining food-label scanning, health analysis, personalized recommendations, and consumer-compliance assistance.

Objectives and Research Gap

Existing food-scanning applications mainly focus on nutritional information, healthier product alternatives, or personalized dietary recommendations. Other research addresses OCR-based text extraction and food-label claim verification separately.

FoodTrust AI addresses this gap by integrating these functionalities into a single mobile application. Its primary contributions include:

• Automated food-label text extraction using OCR.

• Personalized food analysis and health scoring.

• AI-assisted ingredient explanations and recommendations.

• Verification of product claims against available label information.

• Package compliance checking for MRP, net quantity, dates, and manufacturer details.

• Consumer reporting assistance through confirmation-based access to government reporting websites.

System Methodology

FoodTrust AI follows an integrated workflow consisting of the following stages:

1. Image Capture: CameraX captures packaged-food label images using a smartphone.

2. Text Extraction: Google ML Kit Text Recognition extracts readable text from the images.

3. Information Processing: Java-based modules organize the extracted text into ingredients, nutritional information, and allergens.

4. Profile Management: Users select their own or family-member profiles containing supported health conditions, allergies, intolerances, and dietary preferences.

5. Health Analysis: Java-based rules contribute to the health score, while the Gemini API generates simplified explanations, ingredient warnings, possible health effects, and personalized recommendations.

6. Claim Verification: The application compares product claims with available ingredient and nutritional information to identify potential inconsistencies.

7. Package Compliance Checking: Selected package declarations, including MRP, net quantity, date information, and manufacturer or packer details, are examined.

8. Consumer Reporting Assistance: When a possible issue is detected, the application prepares a report and requests user confirmation before opening the appropriate government reporting website.

System Architecture and Technologies

The application consists of two major modules:

1. Food Analysis Module: Handles image capture, OCR, information processing, health scoring, personalized analysis, and recommendations.

2. Consumer Compliance Module: Includes Claim Verification, Package Compliance Checking, and Consumer Reporting Assistance.

The major technologies used are:

Results and Discussion

FoodTrust AI has been successfully implemented and tested on selected real packaged-food products. Preliminary functional testing demonstrated that the application can capture food-label images, extract and categorize text, generate health scores and personalized recommendations, perform claim and package checks, and prepare reports for user confirmation.

However, quantitative performance evaluation has not yet been conducted. The accuracy of OCR, consistency of health scores, quality of AI-generated recommendations, and reliability of compliance detection remain to be established.

Future evaluation should include:

• Comparing OCR results with manually verified text.

• Assessing health scores against predefined nutritional criteria.

• Evaluating recommendation quality and consistency.

• Testing claim verification and package compliance against labelled reference datasets.

• Measuring precision, recall, and F1-score for compliance detection.

Conclusion

FoodTrustAIisanAndroidapplicationdesignedtomake packaged-food information easier to understand. It uses CameraX and Google ML Kit Text Recognition to capture andextractlabeltext,Java-basedrulestocontributetohealth scoring, and the Gemini API to help explain the results. Firebase Realtime Database supports user and family-member profile management, allowing the application to tailor results to the selected profile. The application also includes Claim Verification, Package Compliance Checking, and Consumer Reporting Assistance.Thesefeatureshelpusersidentifypossibleissues withproductclaimsorpackagedeclarations,prepareareport document,andopenarelevantgovernmentreportingwebsite afterconfirmation.Openingthewebsitedoesnotitselfsubmit an official complaint. The main contribution is the integration of personalized food analysis and consumer-compliance assistance in one application using established technologies. Quantitative evaluation of OCR accuracy, health-score consistency, recommendation quality, and compliance detection remains to be completed, and further testing with verified reference data is needed to assess the reliability and usefulness of the system.

References

[1] E. Dunford et al., “FoodSwitch: A mobile phone app to enableconsumers to make healthier food choices and crowdsourcing ofnationalfoodcompositiondata,”JMIRmHealthanduHealth,vol.2,no. 3, Art. no. e37, 2014, doi: 10.2196/mhealth.3230. [2] C.O.C.Werle,C.Gauthier,A.P.Yamim,andF.Bally,“Howafoodscanner app influences healthy food choice,” Appetite, vol. 200, Art.no. 107571, 2024, doi: 10.1016/j.appet.2024.107571. [3] C. Cornudet, M.-E. Laporte, F. Berger-Remy, B. Parguel, and J.-L.Richet, “When food scanner apps outperform front-of-pack nutritionlabels:Aconditionalprocessmodeltofosterhealthierfoodchoicesintimes of growing distrust,” Psychology & Marketing, vol. 42, no. 8,pp.1963–1978,2025,doi:10.1002/mar.22214. [4] P.SoniandK.Kaur,“ExaminingclaimsonfoodpackagesinIndia:Are they inadequate and deceptive?” Measurement: Food, vol. 11,Art. no. 100100, 2023, doi: 10.1016/j.meafoo.2023.100100. [5] M. Nagayi, A. Khan, T. Frank, R. Swart, and C. Nyirenda,“EvaluatingOCRperformanceonfoodpackaginglabelsinSouthAfrica,” arXiv preprint arXiv:2510.03570, 2025. [6] N. Nag, V. Pandey, and R. Jain, “Live personalized nutritionrecommendationengine,” inProc.2ndInt.WorkshopMultimediaforPersonalHealth and Health Care(MMHealth), 2017, pp. 61–68, doi:10.1145/3132635.3132643. [7] A.Prasadetal.,“NutritionWise:Convertingnutritionallabelsintopersonalized, AI-driven nutritional insights,” Procedia ComputerScience, vol. 283, pp. 3307–3317, 2026, doi:10.1016/j.procs.2026.06.394. [8] Department of Consumer Affairs, Government of India, “LegalMetrology.” [Online]. Available:https://consumeraffairs.gov.in/index.php/pages/legal-metrology-overview. [Accessed: Aug. 30, 2026]. [9] FoodSafetyandStandardsAuthorityofIndia,“FSSAIempowersconsumers to report misleading claims on food products,” PressRelease, Apr. 30, 2025. [Online]. Available:https://fssai.gov.in/upload/uploadfiles/files/Press %20Release_ Misleading%20Claims.pdf.

Copyright

Copyright © 2026 Shivani Bulle, Tripti Inwate, Alisha Memon , Khushi Patil. 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.