Meet Vincent: Computer Vision Engineer at Vizzia

Publié le
min de lecture

Hello Vincent, can you introduce yourself?

Hello, my name is Vincent GUÉRIN, and I joined Vizzia nearly 2 years ago - in November 2021. At Vizzia, I'm a Computer Vision Engineer, working on the algorithm for detecting illegal dumping. I started out with a degree in Mechatronics (a combination of electronics and mechanics), then, with my various experiences, I moved into IT until I joined the Altran Group - in consulting - where I worked for Renault. There, I worked mainly on autonomous vehicles, in particular to locate them on the road and develop their autonomy. Then I joined Vizzia.

Can you describe your role in 3 words?

I would say:

> Research : we are constantly looking for new solutions to improve our algorithm and we are setting up processes to develop a new idea, like all the pipes to be created for example. We also do a lot of Infra: managing the systems that allow the software to work.

> Integration : as soon as we have a new idea and we validate it, we must immediately integrate it into production, because it will potentially be of added value for our customers.

> Communication : being at the very beginning of projects, we are obliged to always communicate about what we do in the team, to share and know the progress and news of everyone.

Speaking of news, what's new at the moment?

At the moment, I'm working on two big projects, or even three.

The first is the analysis of false detections. In other words, we detect a very sufficient level of dumping, but we sometimes detect things that are not illegal dumping. The aim is then to understand what these detections are, why they are detected and how they can be dealt with. We've now started the phase where we're trying to understand why these things are being detected. The aim is to maintain a high level of performance and quality.

The second is when we have new ideas or new functions to implement. To do this, we have created an internal process to ensure that, if we add a new feature, we don't degrade the results obtained previously: that the camera detection rate doesn't decrease, or even increases.

Finally, the third project is to look for a new, more efficient platform for labelling our results, which will save us and our customers time.

Have you had a typical day since joining Vizzia?

A typical day, I can tell you, always starts with a coffee. If there's no coffee, there's no start to the day (laughs). Coffee and then we have little rituals like a meeting with the whole IA team to find out what everyone did yesterday, what we're going to do today, and if there are any blockages that we can resolve as quickly as possible. Because there's no point getting stuck on something for a week and then on Friday saying "I couldn't do it". After that, the rest of the day really varies from one day to the next depending on the sprint phase (a short project, generally lasting a week).

What is your professional challenge?

Our main challenge is to ensure that we have an algorithm that detects as many deposits with as few defects as possible. That's THE ultimate challenge.

Is this ultimate challenge what drew you to Vizzia?

In part, but I think, like everyone else here, that what drew me here was the fact that it's a great project. It's something that makes sense and it's rare in Tech, and in the field of image processing, to find a job that has an impact on cities and the environment. It's also a fast-growing project and I was looking to get away from the highly procedural environment of large companies, which take a long time to validate things, aren't very flexible and where communication is sometimes complicated. So far, it's been nothing but fun. Obviously, the schedule is busier than before, but you're always doing new things and that's great. What's more, we have a great team atmosphere, which is really good. You're happy to come to work, there's no pressure, we're working on topical projects, and at the moment we all share the same energy and the same desire to move forward.

Want some
know more?
Ask us your questions