by Jade James
Griffeye are currently offering a free 90-day trial of their flagship product, Analyze DI Pro, which includes access to the user portal and all the resources within it.
Analyze DI Pro comes with many benefits. You can import all types of data, including everything from CCTV to native forensic image formats, as well as data extracted in VICs. It is possible to start analysing the data straightaway: there is no need to wait for all the processing to finish. The Griffeye Intelligence database also allows you to connect to multiple shared databases to share information with other investigators internationally. The robust image and video hashing can save you valuable time by pre-categorising known data and stacking duplicates.The Face Detection and Recognition feature of Analyze DI Pro finds instances of faces in images and videos using Luxand FACE Technology. With this feature, you are able to track faces and facial features, as well as recognising gender, age and facial expressions.
Notably, one of the most enhanced features of Griffeye Analyze DI Pro is the Artificial Intelligence Griffeye Brain, which automatically classifies child sexual abuse content and other categories of images. It is available as a free add-on to Analyze DI Pro.
For more of a technical overview of Griffeye Analyze DI Pro, you can refer to my previous review (note that this was a review of a previous version of Analyze DI Pro, so some features might be different in the latest version).
With Griffeye Brain, Griffeye aim to innovate how AI is applied to law enforcement work processes. At the moment, Griffeye Brain is only available as a free add-on to Analyze DI Pro. The technology uses an Artificial Neural Network (ANN) to classify images.
There are many crucial benefits of Griffeye Brain, including the ability to filter out relevant data. The tool can automatically detect and classify various objects in large data sets, as well as identify and flag images that may contain child sexual abuse. Using Griffeye Brain can help you to make more informed decisions within your investigations and help prioritise workloads, as all of the initial heavy lifting has been done for you, instead of having to manually categorise images yourself.
Once Griffeye Brain has filtered the relevant material for you, resources for victim or suspect identification can be freed up and redeployed elsewhere. The focus can be placed on images and videos that actually require dedicated attention, which therefore speeds up victim identification.
Another useful side effect of Griffeye Brain is that using the AI to sift through illicit images reduces exposure to material which may cause stress to human examiners.
Griffeye Brain has two core solutions: Child Sexual Abuse (CSA) and Objects.
Griffeye Brain CSA will scan through the data it is given, and bookmark images that it believes to contain CSA content. The AI will then compute a probability score which can be used by an investigator to work out if the file is relevant to the case or not. The Griffeye Brain CSA classifier was initially trained on real CSA case data at Taskforce Argos, Queensland Police in Australia, but has now also been further trained on quality-assured datasets at other police agencies in Australia, Europe and the US. It will only review non-categorised images: those which have not previously been given a category. It is also only functional with images at the moment; the classifier cannot be used on video files. The Griffeye Brain CSA can now run on GPU and CPU; when it was initially launched as a beta version, it was only available as CPU, but as it is now GPU-accelerated, it runs up to five times faster than before.
Griffeye will process a case as normal first, performing all the usual checks against the databases which you have imported. After this, the classifier will then look at all the images that did not match any of the databases. The images will be assigned a probability score between 1-99% (it can never give 100% certainty). Bookmarks will then be created based on the probability score.
Griffeye Brain Objects will automatically detect various objects in images and videos, and bookmark them into different classes. The Objects classifier now has the capacity to identify multiple objects in an image or video, making it easier to build relationships in cases. This classifier runs on GPU only, as it is quite powerful: you need a lot of muscle!
Requirements For Griffeye Brain
• NVIDIA Graphics card
• CUDA Support
• The latest drivers
• 4Griffeye Brain GDDR Memory
Recommended Minimum Specification
• GeForce GTX 1080 Ti
• GeForce RTX 2070
• 8Griffeye Brain GDDR Memory
To run Griffeye Brain, you firstly need to visit the Griffeye Portal and download the plugin for either Windows 7, Windows 8.1 or Windows 10. Then you will need to go to the Analyze Forensic Market in Griffeye and install each individual module. Once you have installed the modules, you can then go to Settings> Plugins and check to see that they are listed.
First create a new case and navigate to the ‘Add Files’ option. You will need to go to the ‘Apps’ tab in this prompt and you will see ‘Installed Plugins.’ Select which Brain modules you wish to run and click ‘Start.’ Depending on the size of your dataset, this may take some time.
Results can be viewed by clicking on ‘Bookmarks,’ which is situated at the bottom right of the screen. For Griffeye Brain CSA, you will have three sets of bookmarks:
• CSA – High (probability score of 95-99%)
• CSA – Low (probability score of 1-5%)
• CSA – Uncertain (probability score of 6-94%)
The bookmarks for objects are assigned based on what is identified within the images. Images can have multiple bookmarks, since a variety of objects can be found. The bookmarks for Griffeye Brain Objects are very comprehensive; not all of the bookmarks will be relevant to every investigation.
Using the bookmarks, you can preview and navigate through the images and perform other useful tasks, such as showing files in the same folder as the image you have selected; showing files with the same EXIF number; within the same GPS range; within a specified date range; and visually similar images, among other options.
With the release of Analyze DI Pro 19.2, the AI Technology Griffeye Brain has been significantly upgraded, with enhanced speed and accuracy, as well as a full version of the Objects classifier. It is now possible to detect multiple objects in an image, which was not the case in previous versions.
Another new handy feature of Analyze DI Pro 19.2 comes with the classification panel: as well as the file count, it now shows the number of stacked photos. Attempts have been made to improve reporting as well, in regards to thumbnail stacked images. Previously when an image had more than one file in the thumbnail stack, you had to go to the ‘Info’ panel and select the ‘Files’ to see a list of all the files that existed in your evidence, spilt in to visual copy groups. Now you are able to simply right click and select ‘Prioritise in file list report’; then you can create a report of this prioritised file and exclude all the visual and binary copies.
In previous releases, you only had the option to filter dates for EXIF data, whereas now you can filter by created date, last accessed date, and last write time. With the date filter, you are also now able to customise the date range using an interactive timeline.
Griffeye Analyze DI Pro is a very useful tool to all investigators who have to process and categorise large quantities of images and videos.
Griffeye Brain is now able to automatically bookmark images which may contain CSA or objects which may be relevant to your case. The only limitations it has at the moment are that it can only run on files that are larger than 100×100 pixels and that the Griffeye Brain CSA can only be run on images. Although the enhanced capability of the built-in AI is very attractive, it is important to remember that using it should only be a starting point for your investigation, and it is not to be solely relied upon. However, as a time saving measure, and also when it comes to investigators’ stress levels, Griffeye Brain is an excellent addition to an investigator’s toolkit.
About The Reviewer
Jade James BSc (Hons) is currently a Cyber Security and Forensics Postgraduate Student. She has previous professional digital forensic experience from working at the UK’s Serious Fraud Office, IntaForensics, the Home Office Centre for Applied Science and Technology, and the City of London Police. Jade has gained experience from conducting computer and mobile device examinations as well as drone forensics, and has been involved with ISO 17025 and Quality Standards both as a Digital Forensic Practitioner and Quality Manager.