PhotoILike
Predicting the most impactful picture of your property via Artificial Intelligence.
This month is European proptech month at proptechaweek. We will narrow our search to proptech startups located in the European union (sorry UK, better luck in December). The first stop we make is in the Iberian Peninsula where PhotoILike was founded.
What?
PhotoILike helps you pick the most appealing pictures of your property. You can upload a large dataset of pictures and PhotoILike will sort them from most to least attractive. Computer vision, as already mentioned in earlier editions of proptechaweek, has advanced very steadily in the past years, but detecting a sheep or dog in an image is still somewhat different from interpreting aesthetic indicators. Aaand that is just that which PhotoILike claims to do… But there is more, the prototype available is based on a standard model of aesthetic, but the vision is that it could be modified to target a more niche aesthetic eg. the aesthetic of 25-year-olds who listen to Goth music.
imagery by PhotoILike
Why
High-quality images are one of the hallmarks of high performing real estate ads. So it makes sense that having an objective and predictable way to select the best shot can give a realtor a competitive edge. Categorizing and arranging big photo sets is labor-intensive, and having a way to remove low-quality shots without human intervention could potentially remove a lot of tedious work that instead can be used creating instead of curating.
Apparently, this AI really doesn’t like structural work.
Why not?
Even though the advantages seem clear, and the technology is cool, the product as of now feels more like a feature than a business. I would love to see this incorporated into a clear solution to an existing problem or added to a real estate marketplace. As luck would have it, that is definitely also a possibility as they provide API access to the service.
The solution as it stands is also very generic and it’s not very clear what the aesthetic is that is targeted. In marketing this is a deadly sin, as the saying goes: If you try to appeal to everyone, you will appeal to no one.
Both of the problems above are probably due to the fact that Photoilike is a very young startup and they are still trying to define the exact problem that they want to solve.
Who?
Photoilike was founded by Manuel Perez and Juan Romero in 2019. It came to be with the help of the Ignicia programme a specific regional support grant which aims to help research institutions located in Galicia to transfer research results to the market through a proof of concept. In fact, Photoilike is a spin-off aimed at the real estate vertical which in turn is a spin-off of A Coruña University.
Who else?
Everypixel, a search engine for stock photos has something similar: Everypixel Aesthetics. It let you grade how awesome your photo is by using AI and it’s pretty awesome itself, but also brutally honest.
imagery courtesy everypixel
Further reading?
Article on Forbes: How To Apply AI To Visual Aesthetics
Computational Aesthetics: shall We Let Computers Measure Beauty?
Personalized Aesthetics: Recording the Visual Mind using Machine Learning
Google is using AI to create stunning landscape photos
Things happening
BUILTWORLD published the proptechmap 100 last week… Charter Cities Podcast had @balajis on last week. Definitely worth the listen… Problems in iBuyers paradise?