CODE-AI: Cocoa Origin Detection Enabled by AI

食品欺诈是一个重大问题，迫切需要低成本、简单易用的认证方法。目前，农产品消费者正在呼吁提高食品系统的透明度，以便为他们的购买决策提供信息，并降低购买不安全/非法/不道德/假冒产品的风险。溯源可以通过对供应链数据的深入跟踪来满足这一需求。然而，目前使用的跟踪和跟踪系统容易受到欺诈的影响(因为它们只跟踪包装)或昂贵得令人望而却步(当需要实验室分析时)。在非洲/南美/亚洲种植的咖啡/茶叶/可可等高价值作物分布在世界各地，其供应链受到欺诈问题的影响，在这些全球市场造成损害。更好的可追溯性可以实现伦理和可持续性主张，让公司和政府对他们的承诺负责(如英国2015年《现代奴隶法》第54条)，并以更低的成本实时更准确地衡量生产的社会/环境足迹。这个可行性项目旨在开发代码-AI，一种**人工智能驱动的解决方案**，用于通过扫描作物来识别可可原产地，然后它将成为更大的作物原产地预测工具CropOrigin的一部分。**我们在这个项目中的重点是将尖端机器学习方法的力量与测量可可豆的内在属性结合起来。**最终的应用(‘CropOrigin’)将涉及一种负担得起的手持光谱诊断设备，以扫描豆类/粉末/食品形式的可可。扫描的生物指纹通过网络连接和用户界面输入支持人工智能的预测工具(CODE-AI，可可)，以即时识别作物的地理来源。它将得到农产品独特地理空间和光谱特征数据库的支持；几个供应链的人工智能算法和软件。该项目取得的进展将通过开发一个可供相关控制机构用来识别欺诈行为的可追溯性系统，以及供零售商、进口商和制造商用来验证产品标签，确认其安全性和合规性，直接受益于消费者。将实现该项目的财团包括：Tiwakiki Consulting(LEAD)，一家拥有重要软件、研发和项目管理能力的中小企业；Rothamsted Research是一家专注于作物研究的领先学术机构，通过分包合同与作物健康与保护(CHAP)合作。虽然目前的项目将专注于可可，但该方法有可能应用于许多类型的农产品，如茶叶、咖啡、多叶蔬菜、坚果等。

Food fraud is a significant issue, providing a pressing need for low cost, simple to use authentication methods. Currently, consumers of agri-food products are calling for greater food-system transparency to inform their purchase decisions and reduce the risk of buying unsafe/illegal/unethical/counterfeit products. Traceability can meet this demand through in-depth tracking of supply-chain data. However, track-and-trace systems currently in-use are susceptible to fraud (as they are only tracking packaging) or are prohibitively expensive (when lab-analysis is required).High value crops such as coffee/tea/cocoa grown in Africa/South-America/Asia are distributed worldwide and their supply-chain suffers from fraud issues causing damage in these global markets. Better traceability could realise ethics and sustainability claims, hold companies and governments accountable to their commitments (like Section 54 of the UK Modern Slavery Act 2015) and more accurately measure the social/environmental footprint of production in real-time, and at a lower cost.This feasibility project aims to develop CODE-AI, an **AI-driven solution** for the identification of cocoa origin by 'scanning' the crop, which then will be part of a larger crop origin prediction tool, 'CropOrigin'.**Our focus in this project is to combine the power of cutting-edge machine learning approaches, with the measurement of the intrinsic properties of cocoa beans, to allow the instant determination of geographical origin.**The ultimate application ('CropOrigin') will involve an affordable hand-held spectral diagnostic device to scan cocoa in the form of bean/powder/food product. The scanned biological fingerprint is fed into an AI-enabled prediction tool, (CODE-AI for cocoa) via network connection and user interface for an instant identification of the geographical origin of crops. It will be supported by a database of unique geospatial and spectral signatures of agri-food products; an AI algorithm and software for several supply-chains.The advances made in this project will benefit the consumer directly through the development of a traceability system that can be used by relevant controlling bodies to identify fraud, and by retailers, importers and manufacturers to authenticate product labelling, confirming its safety and compliance.The Consortium who will realise this project include: Tiwakiki Consulting (Lead), an SME with significant software, R&D and project management capabilities; Rothamsted Research, a leading academic institution focused on crop research, and Crop Health and Protection (CHAP) via a subcontract.While the current project will focus on cocoa, the approach has potential to be applied to many types of agricultural produce such as tea, coffee, leafy veg, nuts, etc.