I-Corps: Historical Photo Identification with Crowdsourcing and Automated Face Recognition

I-Corps项目的更广泛影响/商业潜力是开发一种技术，以产生重大的文化和经济价值，包括承认历史上被边缘化群体的贡献。虽然我们最初关注的是拍卖行、评估师和交易商的用户群体，但这项工作也可以应用于画廊、图书馆、档案馆和博物馆（GLAMs）和家谱学会收藏的大量身份不明的照片，这些协会通常人手不足，很大程度上依赖捐赠者和外部研究人员来识别照片。此外，识别工作流程也可以扩展到美国内战时代以外的其他历史时期。最后，这项工作通过一个向学者和公众开放的商业平台，在技术、艺术和历史之间建立了新的跨学科联系。这个I-Corps项目是基于在历史照片中识别未知人物的技术的发展。这是一项具有挑战性的任务，需要广泛的研究人员来完成，包括记者、历史学家、策展人、家谱学家、档案保管员、交易商和收藏家。目前，这些研究人员主要依靠人工调查方法，比如翻看数百页的参考书来寻找潜在的匹配。基于人工智能的面部识别算法和众包为支持这项任务提供了希望，但必须克服误报、偏见和群体思维等关键缺点。这项工作探索了新的人类-人工智能协作技术，有效地和合乎道德地结合人类和人工智能的互补优势，以支持历史人物识别。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The broader impact/commercial potential of this I-Corps project is the development of a technology to generate significant cultural and economic value, including recognizing the contributions of historically marginalized groups. While we initially focus on the user groups of auction houses, appraisers, and dealers, this work can also apply to the vast volume of unidentified photos in the collections of galleries, libraries, archives, and museums (GLAMs) and genealogical societies, which are generally short-staffed and largely rely on donors and outside researchers to identify photos. Further, the identification workflow can also be extended beyond the American Civil War era to other historical time periods. Finally, this work fosters new, interdisciplinary connections between technology, art, and history through a commercial platform made available to both scholars and the general public.This I-Corps project is based on the development of technology identifying unknown people in historical photographs. This is a challenging task performed by a broad range of researchers, including journalists, historians, curators, genealogists, archivists, dealers, and collectors. Currently, these researchers largely rely on manual investigative methods such as paging through hundreds of pages of reference books looking for a potential match. AI-based facial recognition algorithms and crowdsourcing offer promise for supporting this task, but key shortcomings, such as false positives, bias, and groupthink, must be overcome. This work explores novel human-AI collaboration techniques that effectively and ethically combine the complementary strengths of human and artificial intelligence to support historical person identification.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.