1) The crimes themselves are absolutely intolerable to society.
2) The lack of serious resourcing the seriousness of this crime should warrant.
Part of the presentation was specifically on identifying the location a sequence of photos was taken in the hopes that it would narrow down the search radius for the offender. The team that did the work were members of the ICE HERO program -- one of the coolest veteran-to-civilian programs I've ever seen [1][2]. The folks literally spent weeks "driving" around likely areas in Google maps trying to identify the location. Because the photo was taken from an angle that the car wouldn't have seen, they weren't able to use any automation at all and it literally took a human's intuitive understanding of 3-D spaces to eventually figure it out.
However, what was really tough was that the HERO program only exists because it offers an internship that costs very little to DHS -- they simply can't afford a large group of full-time staff.
I urge anybody who can to petition to better fund anti-child abuse programs.
I don't intend to spoil the enthusiasm here, but wouldn't supporting welfare hence be a more effective measure than investing in fancy tech? That latter only increases the deterrants which seem hard to increase any further to begin with.
Disclaimer: I am a parent, and have fostered neglected children.
The trade in the material is unfortunately large. The descriptions of how the systems worked reminded me very much of the kinds of upload/download ratio systems you might find on software and movie/music piracy boards.
I don't know if they accept volunteers for code contributions, but I hope they consider it if they already don't.
I'm a British citizen but looking at spending 2 or 3 months out in Cambodia working on coding projects an exploring a country I've never visited.
Unfortunately, I know Cambodia has had problems with child abuse and child sex tourism in the past. Working on something like you've just described seems like a fantastic way to explore a country and also track down nonces
Europol has a "trace this item" website: https://www.europol.europa.eu/stopchildabuse
I don't mean to view actual child abuse images but similar to the europol thing ie the actual victims deleted from the images.
Why victims are not complaining about these crimes?
This also presumes they're strong enough to speak up in the first place. If you're a victim of this kind of abuse you usually aren't.
None of these pictures have signs of immediate physical violence, but just looking at them and thinking of their possible backstories makes my stomach churn.
I don't know if that's the most effective strategy, but maybe there are other reasons why they can't just dump them all at once.
That card looks like a pokemon card with a Google Chrome logo. Frustratingly familiar.
[1] https://www.google.com/search?tbm=isch&q=yerba%20rosamonte&t...
The most interesting part are the palms, but they are damaged with the photoshop work. If they are what they seem, this particular kind of palms are picky and do not tolerate any frost. A city near the coast probably.
The blue trafic signal is also interesting.
(Could be this photo related with the coca-cola red seats?)
Then they drove to the location and took out the flag
Joe Rogan on it:
https://www.youtube.com/watch?v=AoBK1_ixiro
Some random on it:
Here's a Twitter thread where BBC solved a murder in Cameroon: https://twitter.com/bbcafrica/status/1044186344153583616?lan...
From a single video of the crime, they were able to figure out where, who, and when (!) the crime was committed. (Using Google maps.)
And here's Jeffrey Lewis presenting how they figured out "Where does North Korea Build Its Missile Launchers?" (1h video): https://www.youtube.com/watch?v=E9QbI8Jf4hg
Using "only" North Korean propaganda videos and commercially available satellite images and published memoirs from NK defectors, they were able to located the building where NK is making their TELs.
Does anyone have any other examples?
they triangulated the location of the ISIS execution videos using nothing more than shadows and geographical mounds in the desert
That's an interesting design of swingset; not one I've seen before.
Unlike the bog-standard swingset in the US, which is typically all steel tube, that one looks like it has extruded (aluminum?) verticals and then a tubular horizontal weldment as the top piece. Further, the horizontal looks like it was hot-dip galvanized as a finished unit. The verticals were painted before assembly, not after, because of the bare-metal bolt heads. The fasteners are presumably stainless, since I don't see any rust marks below them. It's got an interesting bearing/bushing at the top of each chain, probably to make it not squeak; that strikes me as a fairly luxe feature.
I couldn't find any similar ones searching around online, although my Google searches don't turn up many European models at all (thanks, Google).
The photo doesn't go up enough to be sure, but it looks like the top bar is an unbroken one-piece section. That would make it pretty bulky and awkward to ship. (The vertical pieces look like they come apart into about 4 ft sections, which means they were likely palletized at one point. Maybe the extrusions are not done in the same place as the rest of the manufacturing..? That's certainly plausible.)
If someone recognized the design/model, I think that would probably be a significant clue towards the photo's location; at the very least, the manufacturer would probably have a fairly good idea of where they normally sell to.
http://igra23.ru/katalog/karuseli-kacheli-balansiry/kacheli-...
It has the swept-bend tubular corner reinforcements, the polyethylene (or some other kind of plastic) panels on the upper corners, etc. The only difference I can see between the product photo and the Europol one is that, in the Europol photo, the top bar seems galvanized rather than painted. But that's the sort of thing that could easily change as a product is marketed.
It appears the company is out of Krasnodar (Краснода́р), located in southern Russia.
[0] https://www.eibe.net/en/swing-kondor-51007201100.html?c=5010...
I think I would if I could trust that it was only being used for something like this & not commercial advertising type projects.
e.g. https://www.reddit.com/r/whatisthisthing/ works great.
https://www.europol.europa.eu/sites/default/files/styles/eur...
over the green square, appears a whitish blur that could easily say Cicoarea in a stylised way (or maybe is just another picture)
https://www.europol.europa.eu/sites/default/files/styles/eur...
Robinia is a fully hardy and widespread species that can be found in almost all Europe and USA, USDA area 1a.
https://www.europol.europa.eu/sites/default/files/styles/eur...
The big trunk in the background is from the same Fabaceae also. Senna could show a similar bark and leaves. Senna siamea from Thailand for example looks similar. There is a sapling of something that looks like a Psidium also and maybe a Schefflera. All tentatively.
https://www.europol.europa.eu/sites/default/files/styles/eur...
If is Fraxinus excelsior (probably) Europe. If Fraxinus pensylvannica: Eastern half of USA or Argentina.
Like « this picture seems to be located in this area/country » and then people from the area could have a look.
I’m pretty sure people would like to help.
Once AI is brought into equation, project executives put all efforts into removing Humans from the chain because AI is cheaper
See YouTube AI based moderation and infractions, it's too difficult and no where to go if AI slaps you
Given that Google actually has the data as well as the skills and computing power to help out Europol, I'd have hoped they'd be jumping at the opportunity to do so. At the least it would make for great PR, but perhaps they're afraid of the possible privacy issues and discussions as well (I can already see the headlines: "Google knows the location of any yard appearing in a photo!").
In any case, it's telling how humans are so good at this stuff and picking up on clues and patterns. It'll still take a while before machines are as good.
Law enforcement and related organizations already use machine learning quite a bit, particularly for image enhancement. Yes, Google does have a lot of images of various locations from a top-down perspective, but that isn’t helpful for accurately determining a location from the images that Europol collects.
Also, keep in mind that Europol is only posting images here when all other means of determining location and identity have been exhausted. The images are usually indoors and don’t contain enough information for ML to be of any significant use. You might be able to narrow it down to a probably country based colors and design patterns, but that’s hardly sufficient and not solid enough evidence to actually do anything.
What do you mean by that?
That is never what I claimed. First, note that I took a pretty (IMO) balanced view and indicated that this is still a hard setting. Second, note that I did indicate that sufficient training (i.e. labeled) data would be required.
This is what was possible in 2016: https://www.theverge.com/2016/2/25/11112594/google-new-deep-...: "The new deep-learning program churns through millions of photos to determine the best match."
Also see project of a fast.ai participant: "Which of the 110 countries a satellite image belongs to?" (point 13 here: https://forums.fast.ai/t/deep-learning-lesson-2-notes/28772)
> (There’s also no such thing as “deep learning”.)
- https://www.deeplearningbook.org/
- https://www.coursera.org/courses?query=deep%20learning
- https://eu.udacity.com/course/intro-to-tensorflow-for-deep-l...
- https://www.edx.org/professional-certificate/ibm-deep-learni...
> Yes, Google does have a lot of images of various locations from a top-down perspective, but that isn’t helpful for accurately determining a location from the images that Europol collects. You might be able to narrow it down to a probably country based colors and design patterns, but that’s hardly sufficient and not solid enough evidence to actually do anything.
Maybe not completely, but again: being able to narrow it down would already be an incredible help, especially for outdoor pictures (which were also shown in the article's video). I never claimed that a model would completely replace the human process.
Also, I find the downvotes (not saying you) on my initial comment to be in pretty bad form. I'm not Jeremy Howard or Andrew Ng, but don't think I was blowing smoke, and work in the area of data science and ML.
