How to detect fake images?

detect fake images
detect fake images

With the Social networks, fake pictures have multiplied. The media share millions of photos, some real, some fake. Their goal: to have the maximum sharing, likes, and retweet.

Verification is becoming more and more difficult. The most incredulous among you can easily fall for it.

To avoid being fooled, here are some practical tips for spotting fake images.

Do a reverse image search on Google

La image search reversed consists of search by image to find similar images on the web. Google has been offering this feature for several years.

To do this, you do not need any software or technical skills.

You just have to right click with your mouse on the image and click on " Search for an image with Google ».

You can also save the image on your computer and import it directly to the search engine.

To do this:

  • Visit Google Images.
  • Click on the camera icon.
  • Click Import Image then Choose File or Browse.
  • Select an image from your computer and click Open.

This method will look for similar images or on the same theme.

If the image in question has been shared several times by several reputable sites, its veracity seems obvious. But the reverse is just as true.

If the image has been shared by sites which disseminate conspiracy theories, which seek buzz, likes and sharing, then its veracity seems to be compromised.

Detect a faked or edited photo

Forensically is a site that offers you to know if a photo has been faked, or items have been added or changed.
To discover a faked photo, Forensically provides you with a variety of tools.

Among which we find:

  • a magnifying glass to reveal the contrasts between the pixels,
  • a clone detector to highlight the duplicated areas on the same image,
  • a metadata function to have the information hidden in the image.

You just have to uplode your file and use the different menus like Error Level Analysis or clone detection to determine whether or not there has been photo retouching.

Here is a demo in video of Forensically possibilities :