Developing computer vision algorithms for ferrous scrap and secondary materials grading using images and videos as input, incorporating them into digital passports for increasing UK usage of domestically generated scrap.

在为期六个月的项目中，Agave Networks团队将开发一种计算机视觉算法，评估黑色金属废料的物理质量特征，并将其纳入其现有的黑色金属二次材料国内市场在线平台。他们使用材料的实时图像创建和使用数字护照，使英国中小企业能够将废料发电机和收集商直接销售给铸造厂，而无需中间人。英国钢铁加工商和中小企业废料场使用低成本或现有的硬件来收集材料的图像，他们的系统创建个人数字护照来验证材料的位置、日期和存在时间，为英国铸造厂提供了一个经过验证的采购工具，消除了额外中介的需要。他们易于使用和高效的解决方案有助于关闭回收循环，并追踪英国回收的金属材料从原产地到熔炉。他们提出的创新是开发一种计算机视觉和机器学习算法，从实时图像中提取回收的黑色金属批次的特征，包括：* %的污染物* %的氧化和降解*件尺寸*潜在的报废来源*大约的批次体积和重量。目前，黑色金属回收的利益相关者目前依赖于视觉检查进行交易目的的材料验证，这容易导致结果的可变性。错误、延误和滥用。其他分级技术，如XRF和涡流磁选机，既不适合，也不高效，不具有成本效益的混合铁材料；因此，很少或没有在整个供应链中使用。根据DEFRA 2021，英国钢铁铸造行业可以在不增加投资的情况下，将国内回收钢的使用量从每年260万吨增加到350万吨。然而，废钢利益相关者之间的数据差异和不匹配，减少了英国国内废钢产能的使用，并使出口更具吸引力，除了可避免的第2类和第3类二氧化碳排放外，每年还造成2.25亿英镑的损失。该项目专注于可持续英国材料和制造业的挑战-通过帮助英国钢铁制造和回收采用更节约资源的解决方案和技术，包括人工智能，计算机视觉和数据通信平台，以：*为回收的黑色材料创建更具弹性的采购供应链。*为闭环金属设计和开发更有效的商业模式*提供回收铁材料从原产地到成品的可追溯性。*减少英国钢铁制造商和回收商在范围2和3方面的二氧化碳当量排放。*通过增加可追溯回收材料的再循环含量，提高英国制造钢铁的国际竞争力。

During the six-month project, the Agave Networks team will develop a computer vision algorithm assessing ferrous scrap's physical quality characteristics to incorporate into their existing online platform for the ferrous secondary materials domestic market. They create and use digital passports using real-time images of the materials to enable UK SME to scrap generators and collectors to market their stock directly to foundries without intermediaries.UK steel processors and SME scrap yards use low-cost or existing hardware to gather images of their materials and their system creates individual digital passports to validate the material's location, date, and time of existence, providing UK foundries with a verified procurement tool, eliminating the need for additional intermediaries.Their easy-to-use and efficient solution helps close the recycling loop and trace UK-recovered metal materials from their origin to the furnace.Their proposed innovation is developing a computer vision and machine learning algorithm that extracts characteristics of recovered ferrous metal batches from real-time images including:* % of contaminant materials* % oxidation and degradation* Pieces dimensions* Potential end-of-life sources* Approximate batch volume and weightCurrently, ferrous recovered stakeholders currently rely on visual inspection for materials verification for trading purposes, which is prone to results' variability, error, delays and abuse. Other grading technologies like XRF and Eddy current magnetic separators are neither suitable nor efficient and non-cost effective for mixed ferrous materials; hence are little or not utilized across the supply chain.According to DEFRA 2021, the UK steel foundry industry can increase domestic recovered steel usage from 2.6 to 3.5 million tonnes p.a. without additional investment. However, data disparities and mismatches between scrap stakeholders, decreasing UK domestic scrap capacity usage and being exporting more attractive, translated into £225M in losses p.a. in addition to avoidable CO2 scope 2 and 3 emissions.This project focuses on the challenge of Sustainable UK Materials and Manufacturing - By helping the UK steel manufacturing and recycling to adopt more resource-efficient solutions and technologies, including artificial intelligence, computer vision and data communication platforms to:* Create more resilient procurement supply chains for recovered ferrous materials.* Design and develop a more efficient business model for closed-loop metals* Provide traceability of recovered ferrous materials from origin to finished product.* Reduce CO2 equivalent emissions on scopes 2 and 3 for UK steel manufacturers and recyclers.* Increase the international competitiveness of UK-manufactured steel by increasing the recycled content of traceable recovered materials.