VICSafer is designed to analyze large volumes of unknown images and videos that police encounter in child sexual abuse investigations. The software is designed to help locate these files that have not yet been observed by law enforcement and recorded in a hash database.

Key Features

Quick Triage: quickly triage large amounts of unknown image and video content to characterize the media as "Safe", "Safe but Contains Children", "Adult Pornography", "Pornography with No Face", or "Possible Child Abuse".

Portability: The application runs in Docker, making it portable. Simply configure the tool to see your input and output directories.

User-Friendly Reports: Generates user-friendly Excel reports that provide all the data needed to quickly review large volumes of files.

Data Labeling: Produces labels for all data, making it useful for local data scientists to work with.

Integration: Easy for local engineers to integrate into other workflow systems.

No-Cost License: Available via no-cost license to law enforcement.

Technical Requirements

Operating Systems: VICSafer runs on Ubuntu 22.04 or 24.04. It also operates on these systems when virtualized using Windows Subsystem for Linux (WSL) and Docker.

Hardware: An NVIDIA GPU(s) is required to run VICSafer at high speed. We recommend that the computer running VICSafer has at least 16GB of RAM.

Skills: Familiarity with the command line in Ubuntu and basic system administration skills are required to use VICSafer effectively. We provide detailed directions for our users. VICSafer does not have a dedicated graphical user interface.

Model and Training Data Availability

We have SAFER models available for Yolov3, Yolov5, and Yolov8 architectures. Additionally, we license our training data and models to vendors and research collaborators. Our models have not been trained on child abuse materials and our training data does not contain child abuse material.

The SAFER model is trained with 23 different classes to identify human faces, estimate gender and age, and detect body parts, sexual actions, bodily fluids, and other objects of interest to child sexual abuse investigators. We release updated models every several montsh to improve the accuracy of one or more of the model's classifiers.

Integration with Detego Global

Detego has a full integration of the Project VIC SAFER model and Safer Viewing officer safety technology. For more information, visit the  Detego Global website.

Contact Us for More Information

Please reach out to us via our "help" button to learn more about the VICSafer technology.