Automate review of child abuse content with biometrics: AIC paper

A new Australian Institute of Criminology (AIC) paper shows how biometric software can help identify perpetrators to link child sexual abuse material (CSAM) without requiring an investigator to review all files manually.

Seven experts from various backgrounds, including universities and law enforcement agencies, presented the paper “Advancing child sexual abuse investigations using biometrics and social network analysis.”

“Trends in the distribution of child sexual abuse material (CSAM) online demonstrate a growing preference by producers and consumers for video and ‘on-demand’ live streams,” notes the research paper. “These trends highlight a growing need for effective tools for analyzing videos in child sexual abuse (CSA) investigations.”

The paper focused on Biometric Analyser and Network Extractor (BANE), a software system developed by the research team capable of extracting and matching biometric attributes, such as voice and face, from a CSAM video database compiled by law enforcement.

Using 445 videos, BANE identified 222 links between them; these included face, voice or both. This method enables investigators to build more complex networks, connecting victims and offenders across videos.

The researchers did not access the CSAM, so they could not verify BANE’s accuracy. The paper concluded with a call for future research to incorporate labeled data in partnership with law enforcement agencies to establish an established ground truth and evaluate accuracy.

“Future research will need to incorporate labeled data with an established ground truth so that accuracy can be evaluated,” the research paper says.

Given this material’s sensitive and illegal nature, the authors also stress that they should conduct further research into these techniques in partnership with law enforcement.

The paper was authored by University of Adelaide’s Russell Brewer and Katie Logos, San Jose State University’s Bryce Westlake, Michigan State University’s Thomas Swearingen and Auren Ross, South Australia Police’s Stephen Patterson, Deakin University’s David Bright and Defence Science and Technology Group’s Dana Michalski.

These experts hope that their research will help law enforcement agencies better combat child sexual abuse material in the future.

In the 2021-22 fiscal year, the ACCCE Child Protection Triage Unit documented an alarming 36,000 reports of child sexual exploitation.

Last year, Australian Federal Police and Monash University formed a partnership to work on an ethically-sourced database to train face biometrics models for use in combating child exploitation.

Article Topics

biometric identifiers  |  biometric software  |  biometrics  |  criminal ID  |  face biometrics  |  forensics  |  law enforcement  |  voice biometrics