Marco, let’s start with a little about you. How did you come to your interest in multimedia forensics and engineering?
My interest in engineering began when I was in high school. I had some older friends who were in university at the faculty of engineering, and they invited me to some workshops where people from the industry showed the students what engineers had achieved through science and technology: the latest developments in mobile communications (the novelty was 3G at the time!), space exploration, and so on. (Funnily enough, I skipped one day of school to go to university!) There I understood that I loved what those people were doing, I wanted to take part in that.
Multimedia forensics came later, when I met Professor Piva at the University of Florence and then Professor Barni at the University of Siena. There are many professors, but some of them are real teachers: their enthusiasm and positive look at reality is contagious. They introduced me to the magical world of image processing and analysis.
Working with digital images is extremely rewarding because you can experiment with real pixels, you do not have to “simulate” anything. The wide spread of multimedia content makes it evident that we must invest in their protection and forensic analysis: pictures are everywhere! So, I started working on watermarking, then rapidly moved to newer “blind” image forensic approaches.
What are the most significant trends you’ve noticed in the past decade of working in this industry, and what challenges do you see these trends presenting?
One significant trend is rather straightforward to see: the more something is used, the more it will be used also for committing crimes and for legal applications. That is why digital audio forensics entered the courts well before digital image and video forensics: digital audio emerged before digital cameras!
Today, we see a great interest in deepfakes since this technology is widespread and can create credible forgeries that can potentially be used for non-consensual pornography, political misinformation, etc.
My personal feeling is that combating fake contents with technology alone is not a winning strategy. We must combine technological tools for fake content detection together with education about these problems: people should stop believing what they see on the web just because they see it. They should think twice and evaluate the trustworthiness of sources – we’re better than computers at that (well, at least as of now!)
As someone with a deep scientific background, what is most important to you about R&D in a vendor context?
It is important to keep in line with reality. Academic researchers sometimes solve problems that will hardly be encountered in real life. This is not wrong at all, they have to do that, it’s their job to look where everyone else would not look. But there must be a tradeoff, otherwise research loses appeal and, eventually, funding. (I’m talking of technological research of course: engineering and related matters.)
I like that Amped Software invests time and money in research: we have collaborations with several universities, we have master’s thesis students, interns, and employees who are focused on research activities, we publish papers, and we attend conferences and workshops. Theses and internships are a great way to attract new minds into the company and test new experimental approaches; indeed, many people joined Amped from the university channel, including myself.
I also like the way Martino Jerian, CEO and Founder of Amped Software, keeps a good balancing act between solving “practical problems” (working on something which is less scientifically challenging but can readily go to market) and “potentially unexploitable research” (that is: try doing something hard, and if it doesn’t work, we’ll have learned something anyway). If a company only lives off low-hanging fruits, it will never be a game changer in front of new challenges.
At Amped, I have two more colleagues holding a Ph.D. in image processing, a mathematician working on validation, and computer science engineers experimenting with cutting-edge enhancement and authentication technologies. We are a very technically oriented company.
What does a typical day look like for you?
During a typical day, I have lots of calls with the research team, but I also keep in close contact with the developers. I also develop some tools myself from time to time. Whenever I can, I read the latest scientific papers to keep track of what’s happening in our field. Besides that, I spend some time in dissemination and training activities, which I also heartily love. I can definitely say my job is variegated and stimulating!
COVID-19 has challenged workers the world over, forcing work-from-home policies and other changes to workflow. How did that impact your own R&D, and how did you and Amped respond to ensure you could still serve customers and their needs?
I personally worked remotely most of the time even before the COVID-19 emergency took place, so my own work was not impacted much. I was used to having video calls, even many of them, every day.
From a company point of view, we did our best to help our users: we started offering temporary work-from-home licenses to allow our users to keep doing their very important jobs while at home, since many of them had our software installed on a workstation in the lab or office. We also started offering digital licenses since our software was previously dongle based only.
And then we started providing online training, which was something we have been wanting to do for quite some time, and like for many companies, COVID-19 just sped up a lot of plans and processes.
Tell us about your work with FORLAB. How does your hands-on experience as an analyst inform your R&D work for Amped?
This is a nice point! My experience as an analyst makes it possible for me to simultaneously keep both the R&D and final user mindsets. It happens quite often that a user requests a feature and then I say “Hey! I would have loved to have it in that and that case I worked on!” And this could help prioritize improvements in our products.
During my FORLAB experience, I also realized that a lot of people out there suffer from the “CSI effect”: they want software to do the impossible, and I can say this because of what customers often asked me to do and the expectations they had. During our Amped training sessions, we always have to make it clear that forensic video enhancement cannot and should not invent information that is not already there!
What’s next for Amped Authenticate and other projects as we finish out 2020 and head into 2021?
Amped Authenticate just received an important innovation this year: we introduced the Shadows tool, which checks the consistency of shadows in an image (but you can also use it on video frames).
The next big improvement we want to add is video authentication. We’re investing a lot of research and development on that, so stay tuned! As for Amped FIVE and Amped Replay, we’re working on making the interface even more user friendly, improving the presentation tools, and more.
Finally, what do you like to do in your spare time?
I live in Tuscany, a wonderful region of Italy, so whenever I can, I go out and have a walk or a bike ride in the countryside. I spend as much time as I can with my wife and little baby (she’s 2 now): she loves learning something new every day, and her laughs and progresses are the most rewarding experience I have in my life.