Artificial Intelligence and Backdrop Research

By Adam Ochs Fleischer 

Artificial intelligence (AI) is changing the world in myriad ways, from revolutionizing industries and transforming business operations, to enabling apps like TikTok to cleverly recommend exactly the type of video content you might enjoy. Though the truly transformative impact of AI on daily life is yet to come, its current influence is already felt in numerous aspects of our everyday life, including voice-activated assistants like Alexa or Siri, semi-autonomous vehicles, and online customer support. It seems inevitable that we will soon be living in a world in which AI is seamlessly integrated into our lives, augmenting our capabilities and shaping a new era of possibilities.

The paradigm shift brought about by AI has not left Civil War photo research untouched. Last month, I was excited to have been given a chance to test a beta version of the new software Backdrop Explorer, created by Dr. Kurt Luther and his team at Virginia Tech. Dr. Luther explains the particulars of the software and how it works in this issue of Military Images (see pages 8-10). In summation, Backdrop Explorer will allow users to upload Civil War-era photographs and match their backdrops (if possible) to other examples in its database, creating “collections” of photos with the same backdrop. Using AI, users can filter backdrops in the database and “teach” the software to show more refined results that may aid in matching a backdrop. Like Dr. Luther’s sister software Civil War Photo Sleuth, Backdrop Explorer gives users the opportunity to include information relevant to the photos they upload, like the medium in which the photograph was taken, the subject’s identity, who the photographer was, etc.

How will this affect backdrop research and this column?

To compare it to the advent of aviation, my research methods thus far are akin to Leonardo da Vinci’s sketches of a flying machine, and Dr. Luther and his team at Virginia Tech have just built a hypersonic jet. Backdrop Explorer will completely remove what has always been a laborious chore: the manual compilation and comparison of backdrops. As a crowdsourced effort, Backdrop Explorer will benefit from an exponentially larger dataset than what I have previously had access to. This, in tandem with the software’s AI-driven ability to organize the data, will undoubtedly result in dozens, perhaps hundreds, of backdrop collections that no one has noticed or had the ability to recognize and organize.

While Backdrop Explorer revolutionizes the process of identifying and organizing backdrops, it’s important to emphasize that the relevance of “Behind the Backdrop” and backdrop scholarship in general remains undiminished. Despite the software’s remarkable capabilities, it cannot replace the need for dedicated study and analysis in understanding the context of identified backdrop collections. Importantly, it cannot infer where a backdrop was used, and by whom. This is, of course, the principal aim of “Behind the Backdrop,” and the information most valuable for practical use by researchers and collectors. To put it simply, Backdrop Explorer will serve as a powerful tool that transforms and enhances the efficiency of the process, yet the interpretive work will remain manual (for now!).

Final thoughts

Backdrop Explorer’s interface is easy to use and intuitive. Like Civil War Photo Sleuth, I have no doubt users will quickly contribute to the database, fostering its growth and establishing it as an invaluable resource. When it becomes public, I enthusiastically recommend that readers try it out for themselves. Dr. Luther and his team are owed a great debt by those of us interested in this topic.

A particularly exciting aspect of the new software is that a photograph may only contain a small portion of a backdrop, or a distorted view of a backdrop, but Backdrop Explorer should be able to successfully match it to other examples. The software will, in many respects, be superior to the human eye in finding corresponding backdrops and uncovering hidden connections.

My suspicion is that Backdrop Explorer will have a logarithmic growth curve with respect to the number of backdrop collections it assembles. In other words, as the database expands, the rate at which new backdrop collections are discovered will rapidly accelerate and then gradually decrease. The new software poses an interesting challenge in that respect, because there was a finite number of backdrops used during the Civil War era. Let the work begin!

Adam Ochs Fleischer is a passionate researcher of Civil War photography and an admitted image “addict.” He began collecting in high school and quickly became obsessed. He lives in Columbus, Ohio.