About Our Software

Cut through time-sensitive & mission-critical
media analysis backlogs using
 
Exponent FacesTM

Exponent FacesTM implements leading-edge commercial facial recognition technology (in use by military and police organizations), directly within X-Ways Forensics, to detect, match and extract faces from photographs and video files. Quickly and accurately identify victims, missing persons and persons of interest within volumes of collected media, including security surveillance video.




Features at a Glance

Accelerate your crimes against children and CASM based investigations using the speed and power of facial recognition technology.
Feature Description
Face Detection Detect and extract one or many faces from any photograph or video files contained within your X-Ways Forensics case file. Faces do not need to be known. By default, identified faces are saved as new child objects to their parent photograph or video file.
Face Extraction Faces that are detected or matched are saved as new child objects to their parent photographs or video files. In the case of videos, frames are first extracted as child objects to the actual video file. At that point, faces are added as child objects to the individual frames. This approach is in keeping with best continuity approaches to discovered evidence and makes it possible to provide granular reporting.
Face Matching Accurately locate and match persons of interest within photograps and video files contained in your X-Ways Forensics case file. All you need is a photo of the person you are looking for and Exponent FacesTM will do the rest!
Enrolling Faces for Matching In order to use face matching features within Exponent FacesTM, comparison photos ("faces") need to be loaded into the program. This process is referred to as enrollment. To make this job easier, you do not need to fuss with locating or creating individual photos (files). Exponent FacesTM makes short work of this for you by detecting and extracting faces from photos containing a group of people.
Photo Lineups Using Exponent FacesTM, you can easily generate photo lineups of faces of persons of interest for reporting purposes. This capability is facilitated through the inherent reporting capabilities of X-Ways Forensics.
Video Frame Extraction Frames from video files can be extracted up to 20 frames per second (at 50 millisecond) intervals, thereby providing the best clarity, resolution and chances of identifying faces.
Video Playback
Within X-Ways Forensics
When video frames are extracted, they can be viewed one at a time within X-Ways Forensics selecting individual frames in the Table pane. This will automatically load the selected frame into the Preview tab below. By using a combination of the UP and DOWN arrow keys, in succession, it is possible to move through an entire video.

Interesting however, by holding down the UP or DOWN arrow key, it is possible to emulate video playback. Using a very low frame extraction rate (e.g., 34-50 milliseconds), slow-motion playback is simulated with the ability to stop at key intervals, and focus in on specific evidence with distinct clarity.

This incidental feature or functionality can only be described as "Cool!" and "Awesome!"

System Requirements

Facial recognition can be a resource intensive process. As a result, high performing computers are recommended.
Requirement Description
X-Ways Forensics Development of this extension was performed with version 20.9. It is therefore recommended that version 20.9 or above is used to run Exponent Faces.TM.
Disk Space Fast storage disks with large caches are strongly recommended. Use solid state drives for best performance. Make sure you also have ample disk space when importing large quantities and sizes of videos and photos into X-Ways Forensics.
Physical Memory (RAM) It's always a good idea when working with multimedia files to have ample physical memory (RAM). This is a very commonplace requirement when conducting forensic processing and examination of videographic and photographic evidence. 64GB is a minimum recommendation in most cases.
CPU and GPU Fast, multiple core and high thread count processors are recommended for optimal performance.

Product FAQs

Here are some of the more common questions and answers that we have received.

Do I have to train the software to recognize faces?

No, absolutely not! Exponent FacesTM implements leading-edge commercial technology that does not require training. This very same technology is currently in use by military.

Does Exponent FacesTM implement open source for it's facial recognition?

No, absolutely Not! Exponent FacesTM implements military-used leading-edge commercial technology that has been in existence for years. This approach to facial recognition can give you the assurance, especially in a judicial process, that the technology used to detect and match faces is validated and adopted throughout the government, military and private security industry, worldwide.

What is the speed and accuracy of face Exponent FacesTM?

Exponent FacesTM varies in performance. Extraction of faces from pictures is considerably faster than video files because frame extraction takes adds a bit more time in processing, albeit in most cases, the difference is negligible. Speed is also affected by the Detection Level, Maximum Faces to extract per photo or frame, and the Match Accuracy settings. Overall, Exponent FacesTM is a robust and top tier forensic technology that performs exceptionally well.

Are there additional fees or hidden costs to use Exponent FacesTM?

No. Your subscription to use Exponent TM and all X-Tensions that come with it are included in your annual license fee. There are no additional costs to use our facial recognition technology.

What kind of computer setup is recommended for best performance?

When you create a case file in X-Ways Forensics (XWF), Exponent FacesTM creates sub-directories where your case file resides and uses them extensively when extracting frames and faces (reading and writing). For this reason, your XWF case file should reside on a fast R/W SSD drive and your evidence should reside on a different drive with similar performance and capacity features.

Let's not forget that a fast GPU is an important ingredient to overall performance, especially when reviewing large volumes of high resolution media files.

What picture and file formats are supported?

Most common pictures and video formats are supported. For pictures, they include .BMP, .JPG and .PNG. For video files, they include .MP4, .MPG, .MOV, .AVI, .FLV and .VOB. Video support is provided by FFmpeg and so the complete list of supported extensions can be found online.

How do I emulate video playback within X-Ways Forensics

In order to get the best resolution and clarity in your media examinations, attempt to extract frames from videos at the lowest frame rate (Frames Per Second) allowable by the source media file. For example, since most devices today record at a minimum of 24 frames per second, then you can configure Exponent FacesTM to extract frames at 42 millisecond intervals. For videos recorded at 30 frames per second, use a frame extraction rate of 35 milliseconds.

The best evidence is the clearer evidence. If your video has fast moving content, then we recommend using the lowest frame rate setting possible.

In most cases, a rate of 50 milliseconds will provide high quality results.

Is it possible to search a video file someplace in the middle?

Great question and the answer is YES! You have the ability to tell Exponent FacesTM to start scanning at any location within the video. You can also specify when (or where) to stop the scan. Cool, right?!

Does the video or picture quality need to be high?

High quality pictures and videos do produce the best results. However, Exponent FacesTM does work with low quality media and can still pull out faces with remarkable accuracy. This includes low lighting and subjects with partially obscurred faces or angled head rotations (left or right) up to 30° off center.

© API Forensics 2024. All rights reserved.
X-Ways Forensics is the trademark and copyright of X-Ways Software Technology AG