Sign In

Communications of the ACM

Home/News/Copycatch/Full Text
ACM TechNews


View as: Print Mobile App Share:
The new software can distinguish an original image (left) from a retouched version.

Researchers in India have created a deep-learning algorithm that can help identify doctored images.


Indraprastha Institute of Information in India professor Mayank Vatsa and colleagues have created a deep-learning algorithm to help people identify doctored images by focusing on specific areas of the human face.

"We look at areas of the face that are most likely to be altered--the skin around the eyes, the mouth, or the forehead," Vatsa says. "In an untouched image, the face shows up as a blend of finely changing textures. Our algorithm looks for unusual smoothness, a sign of retouching."

When randomly presented with 100 facial images, the software correctly identified 87 images as original or retouched, and the incorporation of gender and ethnicity features into the method has improved performance.

Vatsa is one of many researchers worldwide seeking to apply deep learning to a broad spectrum of areas that include computer security, industrial fault detection, medical diagnosis, and financial analysis.

From The Telegraph (India)
View Full Article


Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account