Unstructured media data. It’s both a boon and bane for our modern society. While we can now record, collect, store, and organize life’s moments better than ever before, all that digital archiving is creating a nightmare for certain industries.
Exhibit A: legal professionals dealing with photos and videos during e-discovery.
On the one hand, with so many new ways to capture moments in time—from traffic cams, to personal mobile phone footage, home security systems, and more—the likelihood that someone recorded something important at some point is fairly high, meaning you’re much more likely to have that smoking gun on record than you were ten years ago.
On the other hand, actually finding that smoking gun inside all that media is about as easy as trying to locate a striped, Where’s Waldo shirt on a packed street full of Spanish matadors on Bull Run day.
We won’t go so far as to say the task is impossible, but certainly not easy.
Luckily, with the introduction of AI-based image classification software, attorneys no longer have to manually sift through a mountain of unstructured media footage, just to find Waldo. Instead, you can let the computer do most of that grunt work for you, freeing up your time for more important questions… like, say, “Why do we still have bull runs, anyway?”
Here’s a look at the history of unstructured media data, and the challenges it’s posed to legal image classification.
THE CHALLENGE OF UNSTRUCTURED MEDIA DATA
Readers who have stayed with us throughout the entirety of our AI Legal Revolution journey are already well versed in the woes that unstructured media has posed on the legal industry.
Unlike its counterpart, structured data—which has inherent categorization that defines the data contained within it—unstructured material is raw with no clear programmatic way to identify its content. Not only does it lack organizational structure, but it also comes in a number of different formats, making it extremely difficult to classify and search during discovery—particularly when dealing with unstructured media (as opposed to unstructured text).
What is Unstructured Media?
Whether they know it or not, most people are extremely familiar with unstructured media. These are files that are exchanged, viewed, commented on, liked, and archived every single day. Things like selfies, mobile phone videos, Zoom Calls, and social media.
Unstructured media can also be found outside your own home in the form of CCTV footage, traffic cameras, satellite imagery, jailhouse calls, body camera footage, and so much more.
Which is great… until you consider just how much of it is out there.
Just How Much is Too Much?
In recent years, unstructured data has become the single biggest contributor to the hurdles faced during document review.
Already, this material makes up roughly 80-90% of the content in any given discovery, with unstructured media files making up well over half of that number—and the amount isn’t slowing anytime soon.
According to some experts, unstructured data is on a staggering, 55-60% annual growth rate in the upcoming years, leaving attorneys with the daunting task of finding Waldo with nothing more than a screen and their own two eyes.
The cost in time and labor for such an undertaking is so daunting, that, in the past, attorneys on both sides of the aisle have agreed to exclude unstructured media altogether, just to avoid all that hassle and expense.
Bottom line? You might have everything you need to solve a case… a clear facial image shot of the murderer… the make and model of a getaway car… six, usable frames of an offender breaking windows at a riot… however, unless you have unlimited cash to spend, and an army of reviewers on standby, that information might end up staying right where it is.
AI LEGAL IMAGE CLASSIFICATION: “SIRI, FIND WALDO”
Modern attorneys are stuck between the need for unstructured media evidence, and the inability to process such large quantities on their own.
They need an automated process for finding Waldo—a robot assistant (a la Siri, if you will) that can handle image analytics for legal tasks. One specifically designed to handle image recognition and classification, which can significantly reduce the time and cost it takes to process photos and videos.
That’s why we bring you Exhibit B: the power of AI legal image classification and recognition.
With AI legal image classification and recognition, attorneys no longer have to agree to leave valuable evidence on the table, just because it’s time consuming and costly to review. Legal teams can now easily find and identify thousands of objects within unstructured media, including vehicles, license plate numbers, people, animals, landmarks, weapons, colors, food, art, clothing, and more. This information is scanned, filtered, and organized into searchable results at near real-time speeds, meaning you no longer need an army of reviewers to find exactly what you need—or a bottomless bank account.
However, while all that is super great, what if your problem isn’t, “Where’s Waldo?” but “Who is Waldo?”
Don’t worry. We’ve got something for that, too.
As anyone who’s ever done e-discovery understands, knowing who and what to look for in unstructured media is just as key as knowing where to look. But what if you don’t know what you’re looking for—or whom, for that matter?
Biometric scanners can just as easily be programmed to search content for a specific image, person, or thing, as it can be to find, identify, and categorize unknowable content.
Using contextual image classification and trace markers as a guide, algorithms can detect, find, and classify anything from a face in a crowd to a license plate on a busy street, streamlining the organization of unstructured media exponentially. Helping you track down clues and connect patterns faster than ever before.
AI LEGAL IMAGE CLASSIFICATION: WHAT COMES NEXT?
As society continues to settle into the new rhythm of our digital era, AI image analytics for legal teams will no longer be something that’s “just nice” to have around. With more and more cases relying exclusively on digital evidence, we predict that in the not-so-distant future, legal image classification tools will become absolutely essential to the modern legal process.
Because when it comes to compliance with judicially mandated procedures, accidentally revealing the wrong individual’s personal identifying information (PII) isn’t a problem you want to have.