Verbeter de strijd tegen kinderpornografie

In Fenoz, we focus on making relevant info available to investigators as soon as possible. This way, investigators can start their image forensics and forensic video analysis at the earliest moment.

How do we ensure high-speed processing in CSAM investigations with Fenoz?

1 The architecture
To increase Fenoz’s speed of processing, we selected the most appropriate architecture. The architecture gives our customers the flexibility to increase processing power when needed. The system can be tailored to our customers’ needs. Click here to find out more about Fenoz’s architecture.

2 Artificial intelligence
The use of Artificial Intelligence (AI) in Fenoz further assists in speeding up tasks for users. In all AI developments, ZiuZ will ensure that the forensic investigator remains in control. Click here to find out more about Fenoz’s AI features.

3 Smart processing
To further increase speed, we have introduced tiered processing. For instance, in Fenoz image, the processing of an image can be divided into five tiers:

• Tier 1: exact hash matching using MD5 or SHA1;
• Tier 2: fuzzy hash matching using PhotoDNA;
• Tier 3: extract EXIF information;
• Tier 4: feature calculation (e.g. visual similarity, face, skin detection, logo’s, etc.);
• Tier 5: AI classification.

Tier 1 will need the least computer resources, while tier 5 will need the most. We show investigators initial results as soon as they become available, to save time. We also give hash matching a more prominent role during the import of files from evidence. When we find an exact match of a file we already imported, the processing of that file can stop immediately. As investigators process more cases, they will have to work on fewer and fewer files. This will increase the speed of importing files.

Save the time of CSAM digital forensic investigators with Fenoz

In Fenoz, digital forensic investigators select the tiers that will be executed by default or on demand. This can save considerable time in the processing of a case. The same principles will also apply to other types of evidence like video, audio, and text. To further optimize speed and reliability, in Fenoz we:

• Optimize Video processing for GPUs and scalability,
• Improve invariance to resolution and size,
• Identify and handle Damaged videos,
• Improve user experience specific to videos.

The benefits of the SCS in Fenoz

SCSs operate independently of each other. Information comes in through a message, the system does its work and sends a message back to all other SCSs. The main advantages of the Self Contained Systems are:

• Adjustments are easy to tackle. Having a collection of smaller SCSs results in a centralized business logic. This makes it easier to identify where changes are needed within the relevant SCS.
• Reduced complexity: every SCS is lighter on business logic. This makes testing easier and faster.
• Independent releases of new features. We'll release and update SCSs one by one, easing the process of updating digital forensics features.
• Frequent updates. We expect to be able to release changes more often and as soon as we test them, enabling swift changes.

Fenoz’s Functional modules: elevate digital forensic investigations

We create SCSs to create scalability and make our digital forensic investigation platform robust and reliable. SCSs, yet, are not visible to the users of the system. For the users, the system is a set of modules where each represents a specific group of functions. We call these modules ‘Functional modules’. Several functional modules are developed in Fenoz. These are created to help digital forensic investigators to empower their CSAM investigations. The functional modules in Fenoz are:

1. Team manager (create, change, and delete a team. Add, remove team members; assign and retract roles).
2. Case manager (create, change, delete, open, clone, archive, import and export a case)
3. Match set manager (add, remove, import and export match sets, resolve quality, and more)
4. AI trainer (compose training sets, train classifiers, etc.)
5. Evidence manager (make a forensic copy, extract items, add and remove digital forensic evidence, etc.)
6. Item processor (exact matching and fuzzy matching. Indexing/feature extraction/summary / EXIF. AI classifier, fake detection, face detection/face grouping and matching)
7. Forensic / Investigative Workspaces (classify items. Investigate, prioritize, filter, add, change, and remove victims or perpetrators)
8. Reporting Services (report perpetrator, victim, item, case, etc.)
9. Management/planning dashboard (plan dates, assign cases to teams, present analytics, statistics, etc.)
10. Third-party connectivity (integration of I-Care interface, external classifiers, or functionalities)
11. System administrator (maintain group policies, and storage locations; configure scalability, alerts, scoring modules)

The internet has always been based on scalability. It can process large amounts of data and handle large numbers of users. These technologies help Fenoz grow without limits. No matter how big digital forensic investigation cases get, Fenoz can handle it all.

Smooth CSAM image forensics and forensic video analysis

Fenoz’s browser-based web technology makes Fenoz extremely scalable. The new architecture made it possible to enhance its user interface. Consequently, image forensics and forensic video analysis can run smoother.

Even if it’s a browser-based web architecture, digital forensic investigators don’t need to link their systems to the internet. A web server running the platform can be implemented within the customer’s network. This way, single-user versions can enable a web server on a single computer.

We’re ready for the future of digital forensic investigations

By using browser-based web techniques, we are ready for the future. In case some of our customers will want to move Fenoz to the cloud (private or public), the transition will be smooth.

A benefit of browser-based web technology is that it will make it easier for more users to work together. There is a central server that hosts the application. Everyone with access via a browser to that server can work on cases.

We make use of Artificial Intelligence (AI) to speed up the classification of material. In Fenoz, we are building several AI classifiers. These are text, image, and video classifiers. But also face detection algorithms and file system structure classifiers. Users can also integrate third-party classifiers and custom-specific classifiers.

Make CSAM investigations easier with Fenoz

With advanced AI, we want to make CSAM investigation easier. We aim to improve the accuracy and rate of detection of unknown CSAM by:

• Creating a granular classifier for CSAM using international standards.
• Creating an AI classifier for faces to find similar faces
• Creating the possibility of using the developed classifiers on video
• Install third-party deep fake detection when available

We enable victim identification specialists to search within cases based on visual content.

Process digital forensic investigations efficiently with Fenoz

In most cases, the outcome of an AI classifier is a level of certainty. For instance, the likelihood that an image or video contains CSAM. Users can use this likelihood to sort the imagery. So, they can look at the material with the highest likelihood first.

Within Fenoz, AI is not used to automatically classify material or take over human tasks. AI is a tool that supports CSAM investigation. It empowers users to process digital forensic cases efficiently, leaving the user in control.

Fast CSAM image forensics and forensic video analysis

We use smart filtering to easily combine different filters and reach a desired subset of files. Fenoz displayed the filtering results extremely fast. This makes it possible for our users to create their workflow. So, they can conduct image forensics and video forensic analysis as fast as possible. For example, filtering on all small images first usually results in a set of images that isn’t CSAM material. The user can quickly scroll through the result set to verify that it isn’t CSAM. Then, the user can classify the entire set as non-relevant with one button click. Combining many filters enables these subsets to be even more precise.