Image categorization, now available in Oxygen Forensic® software, is an innovated way to bring in artificial intelligence (AI) that is capable of recognizing photos and videos containing identified information from one of twelve categories. The current categories currently supported are: pornography, extremism, graphic violence, drugs, alcohol, weapons, gambling, child abuse and images and videos deemed to be risqué. This powerful addition is also able to detect documents, currency, IDs and credit cards.How does it work?
Image categorization is based on a neural network that scans the visual composition of an image to approximate the consistency of enough data that matches a particular supported category.
How could it help?
The task of identifying significant images in either a criminal or civil investigation is extremely time-consuming; often comparable to looking for a needle in a haystack. A single case can easily contain thousands of images, many of which are not relevant to the investigation. A manual examination of these images, even within a thumbnail gallery, can take days when the average investigator can only review a few thousand images per hour.
Running image categorization significantly improves the speed of sorting and analyzing visual data on subject’s device. Artificial Intelligence used in our image categorization helps to speed the process of identification when working with images. Eliminate the digital clutter to get the images investigators need. This of course is not replacement for an investigators oversight, all identified images should be verified and validated. After all, the average person has over 600 photos on their smartphone. It is quite typical to have multiple smart devices involved in a typical investigation, the aggregated digital data can be overwhelming without the best tools to weed out unnecessary information.
Image categorization in Oxygen Forensic® Detective.
Image categorization is available within Oxygen Forensic® Detective at no additional charge. You must be running 12.1 to receive this fantastic addition to the industry’s only all in one solution.
Image categorization can be utilized both when importing device data or on an already imported backups or extraction. In both cases, the investigator is able to select categories they would like to search for and also fine-tune the positive “hit” settings by setting identification thresholds in Options/Advanced analytics menu in Oxygen Forensic® Detective. There are four possible threshold settings: low, medium, high (default) and max. The maximum threshold decreases the false positives and detection rate. So, finding the right setting is right at the investigator’s fingertips.
After running the image categorization in Oxygen Forensic® Detective, the number of matching images for each supported category is automatically tagged and identified in both the Key Evidence and Files sections. The investigator can also review the tagged data, manually excluding the false positives and marking the important data as Key Evidence. Be aware, if the investigator chooses to cancel the image categorization, no tags will be assigned and no results will be saved.