Do you have a Google Lens app on your phone? In case you haven’t heard of it, it’s an app that uses Computer Vision to scan and analyse anything. You point your camera at someone’s handbag while passing them on the street – and in 2 seconds you have the full information about the brand and model of the bag, its price and a link to a shop that offers it with a promotional price. That’s Computer Vision in a nutshell. Computer Vision – after Artificial Intelligence – is the recent hot keyword of the tech industry.
While Google creates more and more experiences like this, the real estate industry is still sticking to old and slow ways to deal with buying and selling properties. If you’ve ever been in a position to buy or sell a home, you probably know what it means – a slow, manual process, with mostly… waiting time for someone to reply to your email, print a document or come to your place to assess and value it.
In 2020, during the deep lockdown phase, Saurabh Saxena, a tech entrepreneur based in London, decided to do something about it. His idea was to implement parts of Computer Vision in a process of home valuation – and make it even better than a traditional, “human” way.
Just like Google Lens, his app scans all the available pictures of a property – both interior and exterior, and “looks” at every little detail of it. Is the ceiling tall? Is the garden large? What’s the quality of the furniture?
Is there any sustainable feature visible, like solar panels or thermostats? And then compares it with the homes in the area.
The result is a price recommendation that comes with specific tips: fork out 550 for a solar panel installation and you might increase home’s value by 4,000 instantly. Or “repaint the wall in natural colours to get approximately 25% more offers”. Saurabh’s company, houzen, launched a Computer Vision valuation tool in June 2021, after 5 months of constant engineering and data science work.
We sat down with Saurabh to talk about how computer vision may change real estate for good and if any other industries could follow suit.
Buying a residential property in the UK is commonly considered quite a headache in the UK . What are the common issues investors usually face while purchasing a residential property?
I’d say after speaking to hundreds of investors, it all comes down to 1) time needed to find and manage the investment, as well as 2) Uncertainty of returns. The lack of transparency causes a general lack of certainty in realising returns, which normally happens far in the future. The property market is not like gold or trading stocks, where the value is real and it can be traded instantly. Typically when you buy a property, it’s considered illiquid, and you buy it for a long time. The general lack of data in the property industry creates an opportunity for price negotiation, which elongates the whole process. It creates the opportunity of the pricing being led purely by demand, and the opportunity for buyers to get duped quite easily by brokers or developers. Price setting or price standardization is a general way to make markets more efficient and trusting.
For example, if you were going to plumbing service and there was a website or a plumber that would just tell you the price of the service instantly, you would feel more certain about the service rather than waiting for a plumber to come over and assess the case slowly while being open for negotiation. Um, and consumers will typically move towards, uh, the, you know, the fixed price setting. The problem with variable pricing is that it causes a lot of haggling and it creates room for negotiation. It creates opportunities for good salespeople to make a fool out of uninformed customers. And we see this across the board in the UK, where the lack of data causes a lot of people to not even enter the industry.
And how can technology help anyone looking to buy or sell a property in 2021 in the UK?
About 20 years ago, Rightmove was introduced, and 10 years ago, Zoopla joined the market. At that point, technology was limited to pictures and short descriptions to explain the value of the property. Nowadays, technology has evolved to “reading” pictures. Computer Vision “reads” pictures, by analysing thousands of pixels as well as algorithms in a second. It can combine a property’s interior pictures with floor plans and translate them into meaningful data. So for example, there’s an apartment in Central London posted on Zoopla and we want to find out its real value and quality. The Computer Vision algorithm can gloss through the floor plan and identify that it has two bedrooms, one garden, two bathrooms, and it’s spread over two floors. It can also assess the interior design quality, height of ceilings, state of lawn in the garden, and analyse the general quality of life in the postcode, plus safety on the streets.
By the end, you will be left with a specific price that the Computer Vision algorithm sees as fair for this flat, after analysing all these factors. What’s more, it will also give you a comparison guide: it might find out that this property has a larger size compared with 99% of units in this postcode, and also there’s only 24% of units in the area that have a garden, so this one is indeed a great deal. This set of data can be used to accurately and instantly derive the price per square foot or rent per square foot for any given property in the UK. So the impact of computer vision and general artificial intelligence is going to be improved transparency, and a lot more data analytics around the valuation and pricing, which will lead to quicker and more reliable buying and selling of properties.
How do you expect the technology to support real estate in the next 5 years?
In general, technology has not been implemented very well in real estate to date. The biggest reason is that the developers or brokers who normally use technology, are not very tech-savvy or see new solutions as a threat to their jobs. So the current usage of audio and video tools (like video viewings) is already a massive up for them in this industry. This can continue to automate and make the whole transaction process move faster in the next two to three years. The upcoming technologies include, for example, the use of blockchain. Blockchain is usually implemented whenever there’s a lot of transactions happening – in real estate it could be done through land registries. The market is very fragmented and no broker or developer has a dominant market share, which means that users will easily default to the blockchain. So it has to be owned by the land registry to ensure a full transaction overview.
The second future technology would be an advanced AI in real estate. In the current phase, AI is focused on data collection – right now most of the data in the industry are not even recorded so it’s a huge challenge to make it happen. If data will continue to be collected via video, audio, or through just general software, then over the next five years, the AI will start producing meaningful analysis, and you could start predicting which person will buy, what kind of property, what is their risk appetite and how much money they would or should allocate to real estate and how much money they could make over some time. In the commercial properties field, AI could help to calculate which of the areas to redevelop, which offices or shops should be converted to residential. So to summarise, we could go from analysis into a prediction mode – and do it in a way that brings a lot of certainty to every decision.
You can visit houzen’s page and test the Computer Valuation tool here: www.deals.houzen.co.uk
Find out more about houzen : www.houzen.co.uk