据悉，近日，阿里安全、清华大学联合发起一项通过极少的训练样本锻造“打假AI”的挑战——“ICME-2022 安全AI挑战者计划第九期：小样本商标检测挑战赛”，并在赛题中放入阿迪达斯、耐克、LV、苹果、香奈儿、Gucci、李宁等商标图片，甚至还为顶流“冰墩墩、雪容融”设置赛道，号召全国顶尖AI人才在初赛50张图片、复赛20张图片的情况下，锻造高效精准的AI识别模型，以此对抗侵犯知识产权的行为。下文让小文智能带您一起来了解“打假AI”吧！
（图片来源网络）
去年，阿里安全搭建了超过150万个商标图片、300万个实例的全球最大奢侈品商标库，以此训练的AI打假算法1秒可鉴别20个假大牌logo。在同期举办的AI挑战赛上，阿里安全与全国顶尖AI人才一起沉淀了提升AI识别准确率和安全稳定性的若干算法模型。性能提升后，AI识别真假奢侈品logo可达到1秒200个，训练成本降低了50%。
随着新兴热门品牌的出现，不法商家除了不断用变异词试图逃脱防控，还会让商标logo“小小地”藏在图里，或者遮挡、涂抹、旋转、折叠商标等。阿里安全算法专家仲德介绍：“有些正品商标和某些山寨商标很相似，极易造成混淆，同一个类别有时存在多种完全不一样的子类型，以路易威登为例，英文LV、四花、棋盘格这三类标志都能代表它。”
新一代安全架构核心商标logo检测算法虽在奢侈品logo识别上取得了一定技术进步，面对快速出现的新品牌，如何打造高效识别、成本低的新识别模型，还有不少挑战，对突发风险依然能保持高识别率也是整个AI行业的痛点，而且需要保护的商标类别数量庞大，涉及数十万类之多，如果要进行AI训练，要对每类商标进行大量标注，耗费大量人力。
为此，清华大学和阿里安全联手邀请全国的AI人才解决这一难题，以极少的样本锤炼出最强的AI识别能力，这样“火眼金睛”的AI不仅可用于保护知识产权，还能用于交通违法事件检测、自动驾驶、工业产品缺陷检测等，以科技创新解决产业难题。
以上是小文智能关于“打假AI”的相关报道。小文智能在此呼吁大家不要使用盗版，保护知识产权，抑制盗版是我们每个公民的责任与义务。小文智能致力于用AI技术解放劳动力，做大众认知与认可的好产品，推进AI科技改善生活。自上线以来，不断提升技术、强化功能、优化产品，实现了持续高质量发展，业务量呈几何级增长，目前，全国上千家企业与小文智能达成合作，累计为客户节约运营成本超千万元。小文智能细心聆听客户的意愿，并结合技术专长和资源，为他们提供卓越的智能语音解决方案，从而帮助他们实现企业智能化转型，创造更多价值。
I don’t think the title of your article matches the content lol. Just kidding, mainly because I had some doubts after reading the article.
Your article helped me a lot, is there any more related content? Thanks!
I don’t think the title of your article matches the content lol. Just kidding, mainly because I had some doubts after reading the article.
Your point of view caught my eye and was very interesting. Thanks. I have a question for you.
Your article helped me a lot, is there any more related content? Thanks!
Thanks for sharing. I read many of your blog posts, cool, your blog is very good.
Can you be more specific about the content of your article? After reading it, I still have some doubts. Hope you can help me.
I don’t think the title of your article matches the content lol. Just kidding, mainly because I had some doubts after reading the article.
Your point of view caught my eye and was very interesting. Thanks. I have a question for you.
Your article helped me a lot, is there any more related content? Thanks!
I don’t think the title of your article matches the content lol. Just kidding, mainly because I had some doubts after reading the article.
Can you be more specific about the content of your article? After reading it, I still have some doubts. Hope you can help me.