Purpose is to provide ECCIE community brief overview content based image search technologies, and how these may cause Members to change their behaviors. Or not.
What is Reverse Image, or Content-Based Search?
"Content-based" technologies analyze the actual contents versus the metadata (keywords, tags, and/or descriptions associated with the, web page its on and ‘nearby’ machine processable text). “Content” refer to colors, shapes, textures, outlines of persons, distances between eyes and thousands of other attributes used to match pictures (or picture fragments, cropped or edited pics). People (including of course Jane Provider and Bob Hobby), cars, bottles, couches or any manufactured item, animals, signs; with sufficiently high resolution contents of documents, license plate or registration numbers, and biometric data, etc.) This software is getting a lot more powerful. In the future it will be used in connection with artificial intelligence technologies which utilize boolean search and deductive reasoning capacities.
The “consumer class” of software is mostly free and fairly powerful.
Examples are TinEye, Byo Image Search, GazoPa, RevIMG, IM2GPS, Idée, Google Images, and imgSeek, among hundreds of specialized software.
Industrial strength software (used by military, customs/immigration and in govt intelligence gathering; in private sector in retail and shrinkage-management, by lawyers, bounty hunters, bill collectors, stalkers and the committed) is even more powerful, utilizing biometric tools for content-based matching. Sally Stalker and Creepy Hubert Hobby can, fairly easily, get access to industrial strength software.
Should Members care? Yes.
Any picture that you have ever posted anywhere, can be found by software. In a significant majority of the time, your pictures can and will be found by anyone looking for it AND there is associated metadata that enables the searcher to learn a trove of personal information about you. Or you about them.
What to Do About This?
+ Learn about it.
+ Be aware that its reaching level of ubiquity
+ POST UNIQUE PICS
+ NEVER post a pic that is also posted on your private FB or any other site (Unless you want the world to know you posted same pics here and on FB, etc)
+ GO BACK IN TIME and think about things you may have posted months or years or even decades ago, and understand that these are “indexed” by machine
Any pics that have ever been posted anywhere, ever DO NOT USE THEM here, unless you want all the associated pics and metadata to be revealed to the world. Use unique pics. (Self shots in bathroom mirror personal and community faves!). If you post it here do NOT put same pic on your FB page, even if you "just love it"... If you send it to someone by email or text or otherwise, consider possibility THEY may post it, with/without your consent.
We on the moderator side of the equator come across lots of “breaches of privacy” that cause anywhere from discomfort to lives-turned-up-side-down, which would not have happened BUT FOR this reverse image search technology and how it can be and is being utilized.
Its good to use pictures. Makes Jane Provider look great, lures Bob Hobby and brings good things to life. Just make sure they are unique and NOT the same pics posted elsewhere. Or not unique and posted elsewhere in which case they (and increasingly likely you) will be found.
19Trees
- 19Trees
- 06-14-2013, 04:47 PM
- GneissGuy
- 06-15-2013, 12:00 AM
Many cameras tag images with info about the camera and exposure, maybe even including a serial number for your camera, or something else that might be useful to find other pictures from the same camera. More and more cameras and smart phones put a latitude and longitude on there that will pin it down to which house you're in.
You can remove much of this information from a picture, and some sites filter it out. Some sites filter it out in the pictures they show on their web page, but keep the information in their own databases.
There is software that can analyze a picture, find certain characteristics like noise and bad pixels and then tell you whether two pictures were taken by the same camera. I don't know if anyone is yet scanning the internet and cataloging pics on web pages like this, but I'm sure it's coming.
Facial recognition software that scans web sites is already happening. Some of this stuff is pretty frighteningly good.
All of this kind of stuff is getting worse all the time.
You can remove much of this information from a picture, and some sites filter it out. Some sites filter it out in the pictures they show on their web page, but keep the information in their own databases.
There is software that can analyze a picture, find certain characteristics like noise and bad pixels and then tell you whether two pictures were taken by the same camera. I don't know if anyone is yet scanning the internet and cataloging pics on web pages like this, but I'm sure it's coming.
Facial recognition software that scans web sites is already happening. Some of this stuff is pretty frighteningly good.
All of this kind of stuff is getting worse all the time.