Will You Be Taking AI on Your Next Vacation?
A package tour around the historical sites of the world may still be of appeal to the older generation, but millennials are looking for something different. Attached to their phones, Twitter, Snapchat and Facebook are feeding them the sights of the world as seen by their friends. Naturally, they too wish to share images of unique sites, little known bars and quirky eateries from their next holiday.
For most elements of the travel itinerary, there are comparison portals available: Cheapest flights, value-for-money hotels or a last-minute cab to the airport. But in the booming area of vacation rentals, the choice often seems to fall to just Airbnb.
However, the market for accommodation websites is quite fragmented. Rivals to Airbnb include platforms such as Flipkey and Homeaway. Niche players also appear attempting to offer similar services with a local touch, such as the now defunct VIRTUAL HOTEL platform that targeted Anfield in Liverpool in the United Kingdom. As such it is challenging to get an overview of what is on offer and what a fair price should be, as it is also not uncommon for properties to be advertised on several sites simultaneously but at different prices.
It was this confusion that led to the creation of Holidu, a website to find and compare vacation rentals. Company founders Johannes and Michael Siebers experienced this lack of transparency first hand and decided to tackle it head on. Unlike a hotel comparison site that can link offers via the hotel’s name across providers, vacation rentals can only be reliably linked via their photos. And this is where the challenge begins.
The images used on each website for the same property can be wildly different, taken from different angles and under different lighting conditions. The owner may have taken one set of images, while an agency or property manager may have taken the others. If comparing properties would need to rely on matching images, it seemed that AI would be the tool required to make Holidu work.
Proprietary AI algorithms were developed by the team to sift through over 400 million images, searching for matches. To keep the task more manageable, the images are drawn from just 200 rental providers. The AI thereby recognizes equal images and is also capable of identifying a room taken from various angles. Image matches with a high similarity are used to match offers from different sites automatically. However, if a certain threshold is not met, manual recognition is undertaken to erase any doubt.
The AI also finds its way into learning user preferences and the properties offered. The more travelers use Holidu, the more intuitive it becomes at offering up great deals in dream vacation homes according to your last specifications. If you’re looking for a great deal for your next special vacation, perhaps the AI-powered Holidu might just find the perfect place for you.