创新背景

谷歌在Google Arts & Culture 项目中使用增强现实、3D体验等数字技术开发出在互联网和手机上可以ji体验艺术展览的“口袋画廊”功能，促使艺术名作与数字技术深度结合，极大地扩展了艺术推广度，增加人们与艺术的接触。

创新过程

古斯塔夫·克林姆特是一位奥地利知名象征主义画家。他创办了维也纳分离派，画作特色在于特殊的象征式装饰花纹，在画作中大量使用性爱主题。《金鱼》《黑羽帽》《女人的三个阶段》《阿德勒·布罗赫-鲍尔像》《满足》《莎乐美》等都是他的著名代表作品。画作因为材质、保存手段和一些外界因素，往往会在世界长河中消失或是原样损毁。

Google Arts & Culture在2021年关于古斯塔夫·克林姆特的在线回顾展中利用机器学习技术，与国际知名的克里姆特专家合作，融合先进的数字技术和专家知识，试验模拟克林姆特在画作中可能使用的颜色，完成在“口袋画廊”中将将有照片等资料保存流传的失传画作重新着色，恢复它们彩色的美丽。这是自20世纪中叶画作失传以来，人们第一次能够以他可能使用的颜色体验他的作品。

机器学习是一门多领域交叉学科，涉及概率论、统计学、逼近论、凸分析、算法复杂度理论等多门学科。专门研究计算机怎样模拟或实现人类的学习行为，以获取新的知识或技能，重新组织已有的知识结构使之不断改善自身的性能。它是人工智能核心，是使计算机具有智能的根本途径。机器学习模拟克林姆特的创作习惯，将丢失原本的画作重新着色后，“口袋画廊”会以增强现实和3D形式将他的一些最具标志性的作品带入网上客厅。加上120多个关于他的艺术和个性的故事，来自全球各文化机构的收藏，促使人们在线上就可以全面体验到在展览馆中的欣赏感受。

创新关键点

创新利用机器学习模拟艺术家创作时可能会使用的色彩，交叉数字技术、计算机技术和艺术品修复，将画作重新着色，在数字技术帮助下体现线上全方位展览。

创新价值

利用机器学习将画作重新着色，为艺术再现提供可能，将社会宏观历史和个人艺术历程相结合，帮助更深入研究画家的生平经历、创作历程和当时社会状况，寻找历史长河中的艺术遗珠。

Machine learning recolors the lost painting

Gustav Klimt was a well-known Austrian Symbolist painter. He founded the Vienna Secession, whose paintings featured special symbolic ornamental patterns, and made extensive use of sexual themes in his paintings. "Goldfish", "Black Feather Hat", "The Three Stages of a Woman", "Adler Broch-Bauer Portrait", "Satisfaction", "Salome" and so on are all his famous representative works. Paintings tend to disappear or be destroyed in the world because of materials, preservation methods and some external factors.
Google Arts & Culture used machine learning technology in the 2021 online retrospective about Gustav Klimt, in collaboration with internationally renowned Klimt experts, to integrate advanced digital technologies and expert knowledge, experimenting with simulated colors that Klimt might use in his paintings, and completing the recoloring of lost paintings with photos and other materials in the "Pocket Gallery" to restore their color beauty. For the first time since the loss of paintings in the mid-20th century, people have been able to experience his work in the colors he might have used.
Machine learning is a multi-disciplinary interdisciplinary discipline involving probability theory, statistics, approximation theory, convex analysis, algorithmic complexity theory and other disciplines. Specializes in how computers simulate or implement human learning behavior in order to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their own performance. It is the core of artificial intelligence and is the fundamental way to make computers intelligent. After machine learning simulates Klimt's creative habits and recolors lost paintings, Pocket Gallery brings some of his most iconic works into the online living room in augmented reality and 3D. Coupled with more than 120 stories about his art and personality, and collections from cultural institutions around the world, it is possible to fully experience the appreciation in the exhibition hall online.