Reverse Image Lookup, Paintings, Digitisation, Reuse
July 17, 2013, 15:30 | Centennial Room, Nebraska Union
Once digital images of cultural and heritage material are digitized and placed online, how can we tell if they are copied, disseminated, and reused? This poster explores Reverse Image Lookup (RIL) technologies — usually used to identify unlicensed reuse of commercial photography — to help in assessing the impact of digitized content. We report on a pilot study which tracked a sample of images from The National Gallery, London, to establish where they were reused on other webpages. In doing so, we assessed the current methods available for applying RIL, establishing how useful it can be to the cultural and heritage sectors.
RIL technologies are those which allow you to track and trace image reuse online. The main commercial service, TinEye, available since 2008, finds ‘exact and altered copies of the image you submit, including those that have been cropped, colour adjusted, resized, heavily edited or slightly rotated’ (TinEye, n.d). Since 2007, Google Image Search has also provided a free service which can find similar images across the Internet. Can these tools provide a useful method for tracking reuse of images of paintings once they are placed online? Kousha et al (2010) published a pioneering study which assessed ‘image reuse value’ of academic scientific images. We believe ours is the first systematic study to use RIL to look at digitized heritage content.
We choose two samples of paintings from the National Gallery: all paintings held in Room 34 entitled ‘Great Britain 1750-1850’, containing 26 paintings by 9 artists, just over 1% of their total number holdings (National Gallery, n.d.). We also created a random sample of 6 paintings, from different artistic periods and of varying levels of fame. We analysed the dissemination of these images using TinEye and Google Image Search, using Content Analysis (White and Marsh 2006) to discover the contexts for image reuse.
We then triangulated findings using web access statistics from the National Gallery’s Google Analytics account, and from the commercial ISP analysis firm Hitwise. Our results show that the most popular paintings (by access) are the most commonly used elsewhere, but we also uncover a feedback loop which proves dissemination of images online provides direct traffic back to the National Gallery’s website. Our content analysis also provides a qualitative analysis of types of image reuse, such as commercial art publishers, blogs, reviews, tourism, image collections, encyclopaedias, other museum websites, DVD cover images, and beyond. We demonstrate that type and volume of image reuse is both subject and artist specific.
This study has allowed us to establish what motivates image reuse in a digital environment. We recommend a framework for data collection that could be used by other organisations. However, we also show that there are limitations to the information that can be gleaned from a study of this kind, due to the problematic implementation of the RIL tools which were not designed for this sector.
Acknowledgments
We thank Charlotte Sexton, Melissa Naylor, and Matt Terrington from the National Gallery, and Mike Tovell from Hitwise.