New Life for Old Photos of Málaga: An App Revives Historical Images with Artificial Intelligence

If you're not careful, hours can pass without you realizing it. The Marina Square without a fountain, the first turnkey towers of La Malagueta... or a crowded Rosaleda that has been in need of remodeling for six or seven decades are just a sample of the hypnotic power of black and white, further enhanced by the infinite scroll of a new, inexhaustible format of photos. Encountering all this visual heritage is as easy as entering an app, Panoruma, and having the snapshots assault the screen without even looking for them. You can also refine the selection with keywords and even post the images on social media with just one click. All quickly and easily thanks to artificial intelligence, which for the first time, is applied to the backgrounds of a photographic archive. What's surprising about this project is that it is the final degree work of Alejandro García Peláez, a Computer Engineering student at UMA, who has used his ingenuity to multiply the possibilities of AI. The result is a natural intelligence that has earned the highest grade, honors, and which, along the way, has made this new tool available on the Internet for any user, bringing visual memory up to date.

Panoruma: An app on old photos of Málaga classified by AI

Panoruma: An app on old photos of Málaga classified by AI

Panoruma: An app on old photos of Málaga classified by AI

Panoruma: An app on old photos of Málaga classified by AI

Panoruma: An app on old photos of Málaga classified by AI

1 /

"Panoruma is a digital historical archive explorer that allows you to navigate and discover historical images in an intuitive way," explains the brain behind this project, who breaks down its operation: "What we have done is add an extra layer to the UMA Photographic Archive with AI and create a more accessible interface for any user." Thus, Artificial Intelligence reads and classifies the images from this repository and displays them as a mosaic, similar to the 'wall' of Pinterest or a social network. This first Beta version of the app thus makes available to anyone 3,000 photographs from the more than 32,000 images that the university service has already digitized, most of the Bienvenido-Arenas collection.

"The galleries are updated daily, but the idea is to increase the stock of available photos," says García Peláez, highlighting that this development was possible thanks to the "splendid" previous work of the UMA Photographic Archive staff "who not only preserved this collection but also recovered it and made it available with excellent quality." Thus, the app leverages the thousands of images that are already accessible on the Internet, but which normally have more restricted availability as they are intended for academic realms, researchers, or general inquiries, through search engines that return lists of photos that you have to click on one by one to enlarge. By applying the "layer" of AI, Panoruma allows direct, immediate, and more visual access which "helps to democratize and disseminate" all that heritage, he maintains.

For any collection

As a UMA student, Alejandro García Peláez, who has just graduated and is already working at Google's Cybersecurity Engineering Center in Málaga -specifically in the 'basement', as the research and innovation team of this corporation is colloquially called-, has focused the development of his tool on the university's photographic funds, although he notes that the app is open to any type of already digitized collections. "The Municipal Archive or anyone can use the application so that AI classifies the images and they can be more accessible," explains the computer scientist, adding that this is just a first step and artificial intelligence is opening new possibilities to the world of old photographs, which can also be automatically restored or turned into animations from a still image.

Retouched by Panoruma

Unretouched

In fact, the app developed by García Peláez uses AI to repair scratches, marks, or flaws in the photos, which are 'read' and classified by artificial intelligence to create their own metadata instead of importing the definitions of the documentalists who have archived them. In this way, artificial intelligence recognizes any textual information, objects, and characteristic buildings, and is capable of recognizing small details that are sometimes invisible to the human eye, although it also has its limitations. For example, it cannot distinguish whether an anonymous street or corner is in Málaga or New York. "It could do so by training a specific model or giving examples to one already trained, although the complexity of historical photos is that urbanism and appearance often change with respect to the present," points out this young expert who has used Gemini, Google's AI, to create Panoruma as it has an optimized model for reading the content of images and creating metadata.

While the system is being perfected, the app already allows exploring thousands of images, refining the search with a word, creating personalized albums, marking favorite photos for quick access, and sharing snapshots of yesterday's Málaga on social networks with friends and followers. "Now that it's launched, my main goal is to see the feedback from people and improve the app based on user response," says Alejandro García, who confesses that his goal was for his TFG to be "useful." Therefore, when he discovered the UMA Photographic Archive and spent "like a fool looking at old photos all afternoon," he had it clear. And he set out to apply AI to that powerful visual universe that connects with the present from the past. After a year of work, the result is available to anyone. And it promises to improve it.