Google enlists video ID tools to combat child porn

Regional criminal office (Landeskriminalamt) officials looks through pictures belonging to a person suspected to have purchased child pornography over the internet, during a search in the persons home in this undated handout photo released in the eastern German city of Magdeburg January 9, 2007. REUTERS/Handout/LKA Sachsen-Anhalt/Akte07

SAN FRANCISCO (Reuters) - Google Inc is enlisting the same image-recognition technology the company uses to trace copyright violations on its YouTube video site to fight online child pornography, the company said on Monday.

Google said it is working the National Center for Missing & Exploited Children (NCMEC) of Alexandria, Virginia to help automate and streamline how child protection workers troll through millions of pornographic images to identify victims of abuse.

The project is applying so-called video fingerprinting technology, which Google has been urging media copyright holders to adopt as a means for policing widespread piracy of professionally created video programming on the Web.

A small team of Google engineers have worked for more than a year with federal agencies and NCMEC’s analysts in its Child Victim Identification Program to create software to automate the review of some 13 million pornographic images and videos that analysts at the center previously had to review manually.

The Google technology promises to let analysts more quickly search the center’s video and image databases to identify files that contain images of child pornography victims. Other tools from Google also help analysts quickly review video snippets.

Shumeet Baluja, a research scientist at Google said in a company statement he had recruited a handful of fellow engineers to create the video detection tools as a side project to their day jobs during the course of 2007. He describes the work in a company blog post at

Reporting by Eric Auchard; Editing by Tim Dobbyn