Also available as PDF (figures and tables appear in the PDF).
Food processing supply chains face increasing challenges related to food safety, product perishability, demand uncertainty, and sustainability requirements. Recent advancements in artificial intelligence (AI) have created opportunities to improve decision-making, automation, and operational efficiency across supply chain activities. Consequently, AI is emerging as a key technology for developing smarter and more resilient food systems. Objective: This review aims to examine the current applications of AI in food processing supply chains and evaluate their impact on efficiency, food safety, traceability, and sustainability. It also identifies key challenges and future research directions for AI adoption in the sector. Methods: A systematic review of 25 peer-reviewed studies published between 2015 and 2025 was conducted. Relevant literature was collected from scientific databases focusing on AI applications in food processing and supply chain management. The selected studies were screened based on relevance, quality, and contribution to the field. AI technologies were classified according to their application areas, including procurement, processing, quality control, packaging, logistics, and traceability. The studies were further analyzed to assess technological maturity, operational benefits, sustainability impacts, and implementation challenges. Results: The review revealed that machine learning, deep learning, computer vision, predictive analytics, federated learning, and blockchain-integrated AI systems are increasingly used throughout food processing supply chains. AI applications significantly improve demand forecasting accuracy, product quality assessment, inventory management, predictive maintenance, and supply chain visibility. Computer vision technologies enhance automated inspection and quality assurance processes, while intelligent traceability systems strengthen food safety and transparency. Several studies reported reductions in food waste, operational costs, and resource consumption through AI-driven optimization. AI-based decision-support systems also contribute to faster and more accurate responses to supply chain disruptions. However, widespread implementation remains constrained by data quality issues, high investment requirements, limited technical expertise, infrastructure limitations, and regulatory compliance challenges. The level of adoption varies across supply chain stages, with quality control and traceability demonstrating the highest technological maturity. Conclusion: Artificial intelligence offers substantial potential to enhance the efficiency, sustainability, resilience, and safety of food processing supply chains. Addressing technological and organizational barriers will be critical for achieving wider adoption and maximizing the benefits of AI-driven food supply chain management.
Keywords: Artificial intelligence; Food processing supply chain; Machine learning; Computer vision; Food safety; Trace- ability; Sustainability; Predictive analytics; Block chain; Artificialintelligence; Foodprocessingsupplychain; Machinelearning; Computervision; Foodsafety; Predictiveanalytics; Blockchain
Artificial intelligence (AI) aims to develop computer systemsthatperformtaskstypicallyrequiringintelligenceassociatedwithanaveragehumanbeing’sintellect.Tasksnormallyassociatedwithhavingintelligenceincludelearning,reasoning,perceiving,deciding,andprocessinglanguage.[1][2][3]Recentgrowth inAIadoptionacrossmanyindustries,includingfood, hasoccurredprimarilyduetotheenhancedspeedof computers,improvementsinsensortechnologies,and largeamountsofdigitizeddataavailabletosupport computations.[3][4]CommonAImethodsusedtoaccomplishthisfunctionalityincludedeeplearning(DL) andmachinelearning(ML),whichallowcomputersystemstolearnfromdataovertime,identifycomplexpatterns,andgenerateautomated(machine-generated) accuratepredictions.[4][5]
Foodprocessingsupplychainsarehighlycomplicated andincludeaconnectionbetweenthefarmandthe end-user(consumer)throughaseriesofprocesses,includingprocessing, packaging, storage, distribution andretail. Alargereasonforthecomplexityoffood processingsupplychainsisthatfoodcanspoilvery quickly; as a result, food processing supply chains canhaveveryhighchancesoflosingqualitythrough spoilage,deterioration,contaminationandlowerdemand.AccordingtoFAO[6],therecontinuetobeglobal issuesregardinghowtoensurefoodsafety,reducefood waste, andcreateenvironmentallysustainablefood supplychains.
Traditionalfoodsupplychainmanagementreliesheavilyonanalyzinghistoricaldata,adheringtoestablished protocols, and conducting manual inspections. For decades,thesekindsofsystemshaveworkedforindustrialfoodsystems,buttheyhavedifficultieswith changes that happen in real time, big, diverse data sets,andregulationsthataregrowingtoughertofollow.[5]Becauseofthis,demandforecasting,inventory management,qualitycontrol,andtraceabilityarestill notoperatingwell. Artificialintelligenceisabigpart ofdigitaltransformationinagri-foodsystemssinceit offerssolutionsthatarepredictive,adaptable,andautomated.AIisbeingusedmoreandmoreinfoodprocessingsupplychainsforthingslikeautomatedquality checks,predictivemaintenance,cold-chainmonitoring,developingeco-friendlypackaging,andidentifying contamination.[7][9] EvenifthereisalotofnewresearchonAI,mostofthe studiesthatarealreadyoutthereareeithernotfinished orjustlookatoneapplicationatatimeinsteadofthe wholesupplychain. Thisreviewcriticallyassembles knowledgeregardingAItechnology,applications,benefits, limitations, and prospective research avenues throughouttheentirefoodprocessingsupplychain. Incontrasttopreviousreviews,whichhaveeitherexaminedalimitednumberofapplicationsorfocused ononepartofthefoodprocessingsupplychain,this reviewprovidesanoverviewoftheuseofartificialintelligenceinfoodprocessingthroughouttheentiresupply chain.ThisreviewclassifiesvariousAIapplicationsaccordingtotheirsupplychainstage,technologicalreadinesslevel,andtheireffectsonsustainability,thereby providingusefulinformationtobothresearchersand industryplayers.
We use a structured narrative literature review approachtosynthesizerecentliteratureassociatedwith theapplicationofAIinfoodprocessingsupplychains. We sourced scientific literature from several wellknownelectronicdatabases:GoogleScholar,Scopus, WebofScience,ScienceDirectandSpringerLink. Includedinthissearchwerepeer-reviewedjournalarticles,keyreviewpapers,proceedingsfromacademic conferences,andauthoritativereportspublishedbyindustryscholars,withthemajorityofthesedocuments publishedbetween2015and2025.
Someofthekeysearchtermsusedtodefinetheliteratureincluded‘artificialintelligence’,‘foodprocessing’, ‘supplychain’,‘machinelearning’,‘deeplearning’,‘computervision’,‘foodsafety’,‘traceability’and‘sustainablepackaging’. Weselectedrelevantstudiesbased ontheirpracticalbenefitintherealworld,thequalityofthestudy,andtherelevanceofthepaper(s)to theoverallthemeandpurposeofourliteraturereview concerningfoodprocessingsupplychains.Wedidnot includeanyagriculturalprimaryindustrypaperswhere theliteraturewasspecificonlytothisareaandwhich werenotrelatedtofoodprocessingsupplychains. Theselectedstudieswereclassifiedaccordingto(i)the employedAItechnique,(ii)thephaseofthefoodprocessingsupplychain,(iii)theapplicationdomain,and (iv)thetechnologyreadinesslevel(TRL).Thisclassificationenabledacomparativeandcriticalsynthesis oftrends,maturitylevels,industrypreparedness,and researchshortages.Thisstudyfollowsastructurednarrativereviewapproachwithsystematicscreeningand thematicclassificationoftheliterature.
AIemploysvariousalgorithmsandtechniquestoperformarithmeticcalculationsacrossfoodprocessing supplychains.Severaldifferentmachinelearningtechniquesareoftenemployedtoassistinpredictingconsumerdemandforfood,findinghigh-qualityfoodproducts,determiningwhichsuppliersshouldbeused,and assessinghowmuchriskexists.Commontechniques includelinearregression,decisiontrees,supportvectormachines, andensemblemodels. [5][10]These techniquesprovideinformationonhowdatahasbehavedhistorically,buttheyalsoassistwithpredicting howdatawillbehavegoingforwardbasedonpreviouslyidentifiedtrends.
Theapplicationofdeeplearningtechniques,particularlyConvolutionNeuralNetworks(CNNs),infoodprocessinghasincreasedsubstantiallywiththeadvancementofcomputervisiontechnologies.CNNsarewidely employedfortheanalysisofimagescapturedduring automatedfoodinspectionprocessesduetotheirabilitytoextractcomplexvisualfeatureswithhighaccuracy.[7]CNN-basedcomputervisionsystemscaneffectivelyidentifysurfacedefects,foreigncontaminants, variationsinproductshape, andcolourinconsistenciesthatmaynotbeeasilydetectedthroughconventionalinspectionmethods.Similarly,Zhu[8]reported thatdeeplearningmodelsoutperformtraditionalmachinelearningapproachesinfoodimageanalysisby providinghigherclassificationaccuracyandgreater robustnessundervaryingenvironmentalconditions. TheabilityofCNNstolearnhierarchicalfeaturerepresentationsmakesthemparticularlysuitableforquality controlapplicationsinvolvingcomplexandheterogeneousfoodproducts.
Hybridorcomputer-assistedartificialintelligenceapproachescombinetraditionalexpertknowledgewith data-drivenanalyticaltechniquestoimprovethereliability, interpretability, and accuracy of decisionmakingprocesses.Thesesystemsintegraterule-based reasoningwithmachinelearningalgorithms,thereby leveragingbothhumanexpertiseandcomputational intelligence.[9]emphasizedthathybridAIsystemsare especiallyvaluableinfoodprocessingenvironments whereregulatorycompliance,foodsafety,andproductqualityassurancearecriticalrequirements. Furthermore,thetransparentnatureofhybriddecisionsupport systems enhances explainability and facilitatesstakeholdertrust,makingthemparticularlysuitableforapplicationsthatrequireclearandauditable decision-makingprocesses.
Recent advances in federated learning and privacypreservingartificialintelligencetechnologieshavefurtherexpandedthescopeofAIapplicationsinfoodsupplychains.Unlikeconventionalcentralizedmachine learningapproaches,federatedlearningenablesmultipleorganizationstocollaborativelydeveloppredictive modelswithoutdirectlysharingsensitivedata. [11] demonstratedthatfederatedlearningframeworkscan effectively support food fraud detection and supply chain analytics while maintaining data privacy and confidentiality.Likewise,[12]highlightedthatprivacypreservingAItechnologiesaddressmajorconcernsrelatedtodatasecurity,regulatorycompliance,andintellectual property protection, thereby encouraging greatercollaborationamongstakeholdersacrossfood supplychains. Thesedevelopmentsareexpectedto playasignificantroleinpromotingsecure,scalable, andtrustworthyAIadoptionwithinthefoodindustry.
FoodprocessorscanuseAI-baseddecision-makingsystemstodeterminetheirsuppliers’efficiency,quantify potentialrawmaterialsavailable,andanticipateproblemswithsuppliesduetoseasonalchanges,weather changes, or market conditions. Predictive sourcing analyticstoolsallowfoodprocessorstomakebetter decisionsonmaterialsourcingbasedonboththeirpreviouspurchasesandconditionsexternaltothestage operationsandhelpincreasethelevelofoperational efficiencywhilesimultaneouslyloweringtheriskofbeing reliant upon any one source for supplies, which helpsimproveoverallsupplychainresiliency.[6][24]
AIisbeingappliedinvariousinnovativewayswithin thefoodindustry,andAI’sroleinimprovingprocesses withinthefoodproductionsectorisbyprovidingassistancewithqualitycontrol. Computervisionsystems arefrequentlyusedtorapidlyandautomaticallyassess dimensions,morphology,colour,surfacedefectsand packageintegrityonproductionlines.[7][13]Thistype oftechnologyensuresthattheendresultsaremore consistentandlesssubjectivewhilealsoenhancing speed.
Additionally,morecomplexdeep-learningmodelsallowforhigherinspectionaccuracywithvaryingtypes oflightingconditionsororiginalproductconditions. [8][14]Asanexample,multimodalsensoryfusionprovides an integrated form of the data from an image sensor,aspectsandachemicalsensor,providingan accuratedeterminationoffreshness/quality,compositionandspoilagewithoutdamagingtheitembeing tested.[15][16]
Anymachinethatfailsunexpectedlycanhaveanenormousimpactonproductionschedulesandtheusabilityofaproductandincreasewaste. Predictivemaintenancesystemspoweredbyartificialintelligence(AI) utilizedatarelatedtoenergyconsumption,noise,vibrationandtemperaturetodeterminewhenaparticularpieceofequipmentisatriskforfailureandschedule maintenancepriortothefailure. Industryexamples demonstratethatusingAIwillsignificantlyreducecosts associatedwithmaintenanceanddowntime.
Fooditemswithahighlikelihoodofspoilingrequire preciseestimatesofhowmuchdemandwillexist.AI forecastingmodelsproduceaccuratepredictionsby analyzinghistoricalsalestrends,weatherdataandpreviousmarketingstrategies,alongwithunderstanding how consumers behave. [5] Retailers and manufacturersusingAI-enabledinventorysystemsexperience stock-outs,reduceexcesswastefromoverproduction, andprovideflexiblesupplychainlogistics.[17][18]
Eliminating the risk of a cold-chain failure is critical to providing safe food products, so maintaining accurateandconsistenttemperaturesfrompoint-ofmanufacturetopoint-of-saleisessential–forexample, viacloud-basedsensorsthatmonitortemperatureand humiditylevelchangesinreal-timeinadditiontoestimating remaining shelf life. [19] AI-based delivery schedulingwillincreasethespeedatwhichproductarrivesatitsdestinationwhilesimultaneouslyreducing negativeenvironmentalimpactanddamagingproduct duringtransportation.[20]
Recentresearchindicatesthattheroleartificialintelligence(AI)canplayinproducingintelligent,biodegradableanddynamicfoodpackagingsystemsisbecoming moreprominent.AImodelingaidsinselectingthecorrect materials for making packaging, improving the barrierpropertiesofpackaging,correctlypredicting theshelflifeofproductsanddetermininghowsustainablepackaging. [9]ThecombinationofAIandlifecycleassessment(LCA)toolscanalsohelpunderstand theenvironmentalimpactofproductsandsupportthe goalsofcreatingacirculareconomy.[21]
UsingAI,blockchain,andAItechnologiestogetherallowsforbettertransparencyinthetracingoffoodproducts,thusallowingforeasierdetectionoffoodproductswhichcouldbecontaminatedortomanagethe productrecallprocess.[20]Federatedlearningframeworksenhancetheabilitytoexposefraudulentfood productswithinworldwidedistributedsupplychains wheredataarekeptprivate.[11]Withthegrowinginterestofregulatoryagenciestobetterunderstandand utilizeAIsystemsforfoodsafetypurposes,manyresearchersareapplyingexplainableAImethods(such asLIMEandSHAP)tohelptheseagencies.[22]
TheutilizationofArtificialIntelligence(AI)methodshas thepotentialtoprovidesignificantvaluetocompanies withinthefoodproductionsupplychain.Advantages includelessfoodwaste,increasedoperationalefficiencies,higherproductconsistency,improvedfoodsafety systems,andimprovedsustainabilityresults.[23]UsingautomationpoweredbyAIallowspeopletomake decisionsbeforethingshappen,aswellastocreate strongsupplychainswhenthingsarenotclear.[16]
TherearemanybenefitsthatcouldcomefromusingAI infoodprocessingsupplychains,butthereareseveral obstaclesthatneedtobeovercomefirst.Obstaclesincludehighcoststoimplement,difficultyinconnecting tolegacysystems,ashortageofavailableexperienced personnel,andconcernsabouttransparencyandcompliancewithregulations. [5][6]Small-andmediumsizedfoodbusinessesfacealargeramountofdifficulty inovercomingtheseobstaclesthanlargebusinesses. [19]
Research indicated that the majority of artificial intelligence (AI) applications (quality inspection, predictivemaintenance,anddemandforecasting)have reachedmedium-to-highlevelsoftechnicalreadiness (TRL6–8),whilemostapplicationsrelativetoAI-driven sustainable package development, federated learning,anddigital-twin-basedsupplychainoptimization havelowerTRLs(generally3–5),demonstratingsignificantopportunityforbothtranslationalresearchand industry-basedresearch.[9][23]
Futureinvestigationsshouldfocusoncreatingmodelsofartificialintelligencethatcanprovideexplanationandapplicabilityforfoodsafetyevaluations,lowdata learning methods for small processors, federatedanalyticsforcooperativesupplychainsandoptimizedmodelsthatreflectenvironmentalconsiderations. There is still a major problem in linking the accomplishmentachievedinthelaboratorytolargescalecommercialaccomplishments.
Theuseofartificialintelligenceischangingtheway thatsupplychainsinfoodprocessingoperatebyallowingprocessestobeautomated,data-driven,and predictiveforfutureoccurrences.Althoughmuchadvancementhasbeenmade,thereremainmanytechnical,organizationalandregulatoryobstaclesthatmust beresolvedpriortowidespreaddeployment.Adopting andimplementingAItechnologythoughtfullyandresponsiblywillenablemoreefficient,safe,sustainable andresilientsupplychainsforfoodprocessingaround theworld. Foodprocessors,supplychainmanagers, policymakers,anddevelopersoftechnologythatare developingAI-basedsystemsinthefoodsectorfind theresultsofthisstudyuseful. Fromafoodscience perspective,AIoffersunprecedentedopportunitiesto enhancesafetyassurance,qualityconsistency,andsustainabilityacrossfoodprocessingsupplychains.