Intelligent image categorization with Oxygen Forensic® Detective

The functionImage CategorizationofOxygen Forensic® Detectiveuses artificial intelligence to automatically analyze photographs and videos, classifying them into more than 30 categories of investigative interest, such as documents, drugs, weapons, license plates, credit cards, faces, graphic violence, pornography, vehicles, maps, tattoos, among many others.

Thanks to a neural network trained to recognize visual content, this technology allows you to quickly locate relevant evidence within thousands of multimedia files, significantly reducing the time researchers spend manually reviewing images.

The most recent version incorporates an improved categorization engine, with greater precision in identifying content and support for new categories. Additionally, it can take advantage of both CPU and GPU, achieving considerably higher performance when using graphics processing.

Another of its advantages is the possibility of configuring confidence thresholds (low, medium, high and maximum), selecting the categories that will be analyzed and maintaining a standardized and documented process, facilitating consistent, auditable investigations that are aligned with institutional procedures.

The results are integrated with the Oxygen ecosystem, allowing the sharing of previously validated evidence throughOxygen Review Center, favoring collaboration between investigators, supervisors and other parties involved without compromising the integrity of the evidence.

In investigations that may contain tens of thousands of images, automatic categorization allows us to eliminate much of the irrelevant content and quickly direct attention to the material with the greatest evidentiary value, accelerating forensic analysis and optimizing the work of investigative teams.

How image categorization works in Oxygen Forensic® Detective

The functionImage Categorizationis included at no additional cost inOxygen Forensic® Detectiveand can be used both during the import of data from a device and on an extraction or backup already incorporated into the case.

The researcher can select the categories they want to analyze and configure the sensitivity level using four detection thresholds:Low, Medium, High (default) and Max. The higher the threshold, the lower the number of false positives, although the number of matches detected will also decrease.

Once the analysis is complete, images and videos that match the selected categories are automatically tagged and made available in the sectionsKey EvidenceandFiles, where the expert can review the results, rule out false positives and manually mark the most relevant evidence.

Integration with Oxygen Review Center

Image categorization not only streamlines evidence identification within Oxygen Forensic® Detective, but also integrates withOxygen Review Center, allowing previously validated content to be shared securely with supervisors, investigators, prosecutors or other research participants.

Instead of reviewing thousands of multimedia files, reviewers receive only the images and videos previously classified and selected by the expert, which speeds up decision-making and facilitates collaboration without having to grant access to the entire forensic case.

This workflow allows you to validate results, confirm or rule out false positives, and quickly prioritize the most important evidence, maintaining a consistent, documented, and easily auditable analysis process.

As part of the ecosystemOxygen Ecosystem, the combination ofOxygen Forensic® DetectiveandOxygen Review CenterIt helps optimize the review of digital evidence, improve collaboration between the different actors in an investigation and strengthen the traceability and technical defense of the forensic process.