A Lesson in Consistency With Fulya Lisa Neubert
By Lars - 3 min read
We had EyeEm Vision pick out a photography with particularly great style.
EyeEm Magazine Vol IV. is all about computer vision: How a machine sees the world and understands beauty. To celebrate the new mag, we’ve done an experiment and asked EyeEm Vision to find community members with a particularly consistent style. Then we asked them how they achieve it.
Today we bring you EyeEm’s own Fulya Lisa Neubert. Her selection may seem curious at first: She has helped train EyeEm Vision to recognize beautiful photos. Read on to find out what role Fulya played and how the technology is inspired by her taste.
Who are you and where are you from?
I am a photographer and work as image annotator at EyeEm in Berlin.
Please pick your favorite own photo from EyeEm.
Why did you select this picture?
At the risk of it sounding like a cliché: This is one of my favorite places and I love how it looks imperfect with all the light leaks and grain.
What do you generally like to take pictures of?
Urban landscapes and architecture mostly.
How do you make sure they have a common style?
I usually find myself looking for lines, details and interesting color and geometric combinations. That – I guess – ends up defining my style over time.
What advice do you give other photographers?
What was it like to train EyeEm Vision?
I was one of the few people who collected the data for aesthetics training which is more than 200k photos right now. We went through 200k ~ photos that were uploaded to Eyeem, and put them in categories according to their aesthetics. And the algorithm was trained with these photos and categories. Most images had been correctly judged, but there are some concepts when I disagreed with the machine: For instance, the technology considers minimalistic images a bit boring, while I enjoy their stripped-back looks. And I made sure that film photos weren’t just considered too grainy but judged accurately for how beautiful they are.
You were the corrective element?
In a way, yes. I told the machine how it should think. So no wonder it liked my photos, even though the odds of being picked are still really low.
Want to read more about how computer vision is changing photography? Head over to swag.eyeem.com and get yourself a copy of EyeEm Magazine Vol. IV.