Google Suggesting Photo Location Recommendations
A recently granted Google patent explains how Google may find photo location recommendations. It describes how it might use something like Google Now to recommend “photogenic locations to visit.”
The patent tells us:
The present disclosure relates generally to systems and methods for recommending photogenic locations to visit. More particularly, the present disclosure relates to prompting a mobile device user that a photogenic location is nearby based on clusters of photographs.
When vacationing or visiting an unfamiliar location or a familiar location at an unfamiliar time, a person may desire advice regarding popular sites to visit or landmarks to see. In particular, a person may desire advice regarding an interesting view to see or phenomenon to experience.
It tells us that Tour guide books miss out on places that people like taking photos at because they focus on “tourist-style” landmarks, and that Online Recommendation systems that might identify such spots tend to focus mostly upon businesses such as restaurants, and the patent tells us:
While such systems can be useful for selecting a restaurant in an unfamiliar location, they fail to provide additional, non-commercial knowledge concerning photogenic locations.
.
The invention described in this patent involves prompting a client device when it’s near a photogenic location, that has been identified as a place where people like to take pictures, or “Each of the plurality of photogenic locations can have been identified by clustering a plurality of photographs based on geographic proximity.”
The patent is:
Systems and methods for recommending photogenic locations to visit
Publication number US9014726 B1
Publication date: Apr 21, 2015
Filing date: May 3, 2013
Priority date: May 3, 2013
Inventors: Andrew Foster
Original Assignee: Google Inc.
Abstract:
Systems and methods for recommending photogenic locations to visit are provided. One aspect of the present disclosure is directed to a computer-implemented method for recommending photogenic locations.
The method includes receiving a signal indicative of a geographic location at which a client device is located. The method further includes determining whether the geographic location is within a threshold distance from at least one of a plurality of photogenic locations. Each of the plurality of photogenic locations can have been identified by clustering a plurality of photographs based on geographic proximity. The method includes transmitting a prompt to the client device when the geographic location is within the threshold distance from at least one of the plurality of photogenic locations.
The prompt can indicate the existence of at least one photogenic location that is within the threshold distance from the geographic location.
There are a few different ways that such locations can be identified.
Someone might participate in having their mobile device report their location when they take photos, and it can create a location history profile for them.
Also, photos might be geotagged, and a database of geotagged photos could be analyzed to determine the locations where those images were shot. Geotagged photos “can include EXIF data indicating latitude, longitude, date of capture, and a time of capture.”
Such a recommendation system might make recommendations as to good places to take photos of sunrises and sunsets, based upon the time they were taken and a comparison to the actual time of those events.
If Google makes such a recommendation, and the person being recommended the location may show that they stopped and took the advice of the recommendation, through something like Google’s Auto-backup photo service. If they do, this recommendation system may continue to make recommendations about Photogenic Locations.
In addition to clustering the geographic locations where photos where taken, a clustering algorithm might look at other information, such as “metatags, keywords, text annotations, or comments provided in the context of a sharing or social media platform.”
The patent also discusses the possible use of “visual feature matching”, to help identity what is being photographed, such as a historic courthouse versus a popular artistic sculpture.
Google might also attempt to see what is located at these geographic clusters using its knowledge of the locations of landmarks. at those clusters.
The photo location recommendations patent also tells us about a “global interestingness score” that can be determined for each photogenic cluster. This score might be “determined for each of the plurality of photographs included in a cluster based on one or more signals indicative of online activity associated with such photograph.” So, if a photo is shared from the location on Instagram and it gets positive feedback at Twitter, Facebook, and Instagram as a result, that could help increase that interestingness score. That score could play into whether a recommendation for a spot to take a photo might be made.
Thanks for this Bill.
The big thing about this article for me, is in the last paragraph.
“So, if a photo is shared from the location on Instagram and it gets positive feedback at Twitter, Facebook, and Instagram as a result, that could help increase that interestingness score.”
Its not often that you read about Google openly considering using external social data for use within their algorithm. Don’t you think it will probably just Google+ that’d be primarily used for a Global Interestingness Score?
Thank you for this useful sharing.
“So, if a photo is shared from the location on Instagram and it gets positive feedback at Twitter, Facebook, and Instagram as a result, that could help increase that interestingness score. That score could play into whether a recommendation for a spot to take a photo at might be made.”
Interesting as always Bill! I see a patent like this and the recent deal with Twitter makes sense as you can see value from the data, otherwise, Realtime search was a collosal flop and the deal makes zero sense. I don’t know that I have seen a feature get demoted down the SERP and dropped so fast as that was from Universal Search.
Hi Terry,
Yes, this patent does point out how some social attention data could be used together with Google’s data in ways that could be very beneficial to the search engine in providing services. That kind of interestingness score could apply to other things as well, which hopefully we will see Google take advantage of.
After Apple acquired Topsy, and their social analytics approaches, this seems to be an area that Google needs to become good at, or potentially fall behind in.
You’re welcome, Rinjani
Hi Andrew,
That was really interesting to see!
Google entered into a new partnership with Twitter in their use of Twitter’s firehose of data. Google could be using that data in ways that enable it to calculate an interestingness score like that, in addition to showing real time Tweets.
I can’t imagine that it has taken this long for Google to address the significance of pictures today when the increase in cell phone usage has put cameras in so many people’s hands. Also, photos are such a growing part of SEO. This sounds like it will help a lot by focusing photo takers in the right direction.
Hi Carly,
Google has been active in setting things up like Auto-backup for images. making photos an important part of Google+. and offering things like auto-awesome animation and stories, and editing tools for photos. Building something like this into a recommendation program such as Google Now makes a lot of sense, for the points you raise, too.
This is interesting – This has been a hot topic this morning….From a company perspective I guess posting dynamic photos is important to try to get the image out there.
Nice Post, I like it very much. Thank you for sharing this informative post.
Thanks Bill!
Great sharing, I have one question that how this “global interestingness score” will effect us as a general users on Google search engine or Google+:
Don’t know if it’s just me but this seems like a very odd idea to me…Surely then you’re just going to get the same photos as thousands of people before? Way to go lack of imagination! 😉
Hi Bill,
It’s quite interesting to know that Google is actually opening up for external data given by big social media sites in order to identify and locate places more accurately. Knowing the fact that Google also has its own social media place like Google+ but still considers pictures coming from Twitter, Instagram and Facebook for scoring, it goes to show that they are investing a lot to maximize the potential of this said patent. Well, I guess it does not hurt them to collaborate with these social media havens if their aim is to help netizens, tourists and businessmen in their traveling and promotional endeavors.
Hi Farell John,
It is interesting that Google may use that social media data to come up with a global interestingness score. If they are doing that with the Twitter Firehose of data, I think it’s a good way to use that information – better than just publishing tweets.
Hi Tom,
Chances are there will be some uniqueness to those images, given different photographers, with different cameras and different filters, different levels of sunlight, and filters, and so on. This recommendation system is making recommendations for places for people to take images; I think it’s an interesting idea.
Hi Muhammad,
The global interestingness score seems to be primarily for purposes of making recommendations to people who have a recommendation system such as Google Now on their mobile devices.
Hi..
Thanks for sharing this post really its a very interested. This has been a hot topic this morning….From a company perspective .
Great article!!
worth one time reading. Thank you for this fantastic article
Hi Khamis,
Thank you very much.