Quiztime: How To Clean Pictures For Better Search Results

In last week’s Wednesday Quiz Tilman Wagner posted this photo and asked us several questions about it:

1) Who posted it and where?

2) What was the occasion?

3) Did Tilman participate, and if so, how?

So where to start? We need to know the location before we can answer any of Tilman’s questions. There are a few clues in the image that might make a good starting point: the language on the screen is is in British English (organisation spelled with an ‘s’) and it suggests a networking event or meeting of some kind. There’s a WiFi sign but unfortunately the network name is too small to read, or else we could have had some fun with Wigle. No quick wins so far.

An important rule in geolocation (and investigation generally) is that the more unique or distinctive a feature is, the more useful it is for finding the origin of a photo. By far the most distinctive feature is the unusual pyramid shaped building that it visible from the window. If we can locate this, we can work out where the photo was taken from.

This is the most distinguishing feature in the photo and will be the key to geolocating it.

Cropping the image and conducting a reverse image search seems to be the most obvious way to find the building, but there’s a big problem with that approach in this instance. When we look through a window with human eyes, we subconsciously ignore details like window frames and curtains to focus on what is outside. Unfortunately reverse image search engines aren’t smart enough to do this yet and they don’t distinguish between the foreground noise and the background detail that we are interested in. If you tried to solve this puzzle with this method you probably got a lot of matches that looked like this:

Yandex search results find a lot of matches…but only for window frames.

Reverse image search engines show their limitations here. They focus on the window frame and curtains but don’t focus on the building outside that we’re interested in. So what can we do?

Removing Noise From The Image

What we need is an easy way to remove the foreground clutter from the image so we can conduct a better reverse image search. Cleanup.pictures is very easy to use and makes it easy to remove unwanted features (like window frames) from images. Instead of leaving gaps, the software guesses what should appear in the image based on the rest of the image content and fills in the details. It will be perfect for what we need.

First we upload our cropped image to the website.

Cropped image ready to upload to Cleanup.pictures.

Next we use the eraser tool to highlight the details we want to remove.

Highlight the areas for removal.

Here’s the final result – not too bad!

The result with all the foreground noise removed.

By using the smart eraser Cleanup.pictures removes the unwanted details and fills in the gaps based on the unaltered parts of the image. The result you can see above gives a clutter-free image that will give us a much better chance of success with a reverse image search. Here are the reverse image search results with the foreground clutter removed:

Search results without the foreground noise.

Much better! With the window frame removed, we get much better matches for the building in the background. We learn that it is the National Library of Latvia in Riga. By examining the area in Google Maps, we can see that there is only one building that Tilman’s photo could have been taken from: the Wellton Riverside Spa Hotel. The other buildings opposite the National Library do not have glass fronted balconies like the location in the original picture.

The Wellton Riverside Spa Hotel in Riga.

By way of confirmation, a quick look at the conference facilities on the hotel website brings up a room that matches the one in Tilman’s picture. The National Library is even visible out of the windows.

The conference room at the Wellton Riverside Spa.

So we know the location – how do we find out about the conference?

Finding The Event

Tilman posted this challenge on Wednesday 26th October and the photo was posted on the same day, so it’s reasonable to assume that the event happened in the few days leading up to the 26th. This means a good starting point is to found out what events were held at the Wellton Riverside Spa that week. It’s a simple piece of information but it’s surprisingly tricky to find out. The hotel website doesn’t list upcoming private events and search engines aren’t really too helpful. Eventbrite has a great search feature that allows you to search for events by location, but unfortunately it only works for current and future events, not past ones. It’s much easier to start with an event and find the location, but it seems more difficult the other way around.

So what techniques might help us make the link between an event and a location? Geotagged social media posts are often helpful to tackle this problem. Both Facebook and Instagram have geographic search filters, so they are a good place to start.

Instagram is my preferred resource for this because even though its search features are quite limited, it is still really good for searching photos by location. Simply entering “Wellton Riverside Spa” into the search bar brings up a few locations and hashtags of interest.

All the matching hashtags and location tags for the Wellton Riverside Spa.

One of the geotagged images leads to this image that was posted on 25th October. It refers to the hashtag #MILSummit2022 – so we will need to find out exactly what this summit was.

The image is geotagged at the hotel location.
A second image reveals some of the themes of the event.

The same user posted another photo seemingly taken at the same event. Whatever the MIL summit is, the themes of media literacy, disinformation and verification are very common. We know Tilman works in this field, so this is probably a good indicator we are on the right track. We also learn from the second image that the full name of the conference is the “Baltic MIL summit”.

This post links the MIL event at the hotel to the Ziniu Economic Forum,

By researching the “Baltic MIL Summit” we quickly find that on Monday 24th October the Žinių Ekonomikos Forumas was hosting the MIL Summit at the hotel. The Facebook page for the event contains hundreds of photos of the participants taking part in the conference. Not only do these photos help to confirm the location matches the one in the original photo, but eventually we even find the exact same photo that Tilman used for the quiz. Bingo!

A familiar photo on the ZEF Facebook page.

So we now know that the image was posted by the Ziniu Economic Forum on their own Facebook page, and that the event at the hotel was the Baltic MIL media literacy summit held at the Wellton Riverside in Riga. To finish the quiz we just need to find out whether Tilman took part (if indeed he did).

Linking Tilman To The Conference

Clicking through all 230+ of the event photos was useful because not only did it help to identify the source of the original image, but despite all the attendees being very heavily photographed there was not a single photo with Tilman in it. (If you didn’t know what Tilman looked like already, a little Googling will bring up this bio with a picture.)

Tilman’s picture and bio details.

So there are two possibilities: either Tilman was there and he is very, very camera shy, or he was not there in person at all. On the balance of probabilties, it seems more likely that he did not physically attend the event, but we need to try and verify this first. Are there any other sources of information that might help?

Looking through the images, it’s clear that there are strong links between the Deutsche Welle Akademie (where Tilman works) and the MIL event in Riga. We need to try and explore this connection a little bit more.

A good place to start might be with the conference organisers, Ziniu Economic Forum. Most conferences have at least some publicity and a schedule of events. If we can find these, we might be able to learn a little bit more about the links between Deutsche Welle and the MIL Summit.

The first article on the ZEF website refers to the summit and also includes a schedule for the event (PDF download). There are a lot of speakers – including several from DW Akademie – but Tilman is not one of them.

Day Two of the summit, but no Tilman…

So we know some of Tilman’s colleagues were speaking at the event, and even though we have not found any evidence that he attended in person, the schedule indicates that some of the sessions were streamed over the internet – so perhaps he took part by watching online? Perhaps he did, but it is hard to verify this directly. Can we do any better than just saying Tilman is linked to the conference because his colleagues were speaking there?

We have already made a one-way link from the ZEF to the DW Akademie via the conference photos and speaker schedule, but what if we approach from the other direction and see what DW Akademie have to say in their own event publicity?

By starting at the DW Akademie website we can see that they have two social media news feeds, one for Twitter and one for Facebook. By focusing on the dates of 24-25th October we can see that a live feed of the conference was available (so perhaps this was how Tilman took part?), but the other main publicity feature from DW is a talk by Lena Ritsche on their new game “Go Verify”. Here it is on Facebook:

Facebook mention of Go Verify.

And here on Twitter…

Twitter reference to Go Verify.

The Twitter post contains a direct link to the Go Verify game that is the main focus of Deutsche Welle’s publicity around the Baltic MIL Summit. By following the link and reading about Go Verify we find a familiar name at the bottom of the article:

And there we have the closest link between Tilman and the conference so far. It seems unlikely that he was present, and we can’t say for sure whether he watched online or not – but we can say that the Go Verify game that he helped to develop was the subject of one of the talks.

If you want to learn more about how to use Cleanup.pictures you can wach Technisette and I demonstrate it on David Bombal’s webcast here.

4 thoughts on “Quiztime: How To Clean Pictures For Better Search Results”

  1. Pingback: Quiztime: How To Clean Pictures For Better Search Results – We are OSINTCurio.us

  2. LS
    Het hotel waar hierover gesproken wordt, wat tegenover de bibliotheek zou moeten staan, staat er niet. Gelukkig is het maar een spelletje, groet..

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.