Increasing crime, falling police officer numbers, stubbornly high terrorist alert levels and belligerent state sponsored attacks have kept security in general, and identification in particular, high on the news agenda and current affairs headlines over the past few months. The sheer incompetence of the Russian attempt to murder retired spy Colonel Skripal and his daughter in Salisbury was bad enough and must have caused considerable consternation in the Kremlin at the almost schoolboy level antics of their supposedly highly trained military intelligence agents.
But the killing of the dissident journalist Jamal Koshoggi by a 15 strong assassination squad inside of an embassy of the Saudi Arabian state in Turkey was a complete farrago. Some governments clearly think that they are now above the law, can operate above the accepted norms of community behaviour and this is a most worrying state of affairs. One thinks of the sinking of the Rainbow Warrior by French DGSE agents in New Zealand in July 1985 and the Israeli Mossad assassination of Hamas leader Mahmoud al-Mabhouh in February 2010 as other examples of state sponsored attacks overseas – now more frequent and more belligerent than ever. James Bond, thankfully, is far more professional.
Of course, Super Recognisers were involved in both of these international incidents. Interest in their abilities continues to grow with a team now well established in Germany, interest from South America and the Middle East. It is becoming increasingly clear that the ideal solution to the deployment of modern autonomous facial recognition technology is to combine the science with the skills of the human brain – in particular the enhanced skills which super recognisers can bring to bear at borders, ports and airports, shopping centres and key infrastructure installations.
SRs have appeared in a number of fictional novels, one of them, Violet Hill is reviewed on the website.Allevate, a high technology company have produced a White Paper on the use of FR technology, we particularly like paragraph 3.5.An article about the recognition capabilities of non-SRs begins our news this quarter and reproduced below with our thanks to New Scientist for allowing its publication and highlighting the issue.
Over the autumn I have taken the opportunity of guiding the Association of Super Recognisers to the next level of their development and in shaping the future. I meet regularly with the board and it is a delight to see the organisation grow and develop as more and more countries and organisations discover the benefits of having licenced Super Recognisers within their police and security teams.
As the membership of the Association increases, this growth allows more marketing of the capability to be achieved, creating more awareness of the skills and supporting those who are already committed to making best use of this new and exciting area. I think it most timely to remind everyone who is involved with facial recognition of the Surveillance Camera Commissioner’s ‘Code of Practice’. The Code states, as part of Guiding Principle Number 2, Paragraph 3.2.3 that “Any use of facial recognition or any other biometric characteristic recognition systems need to be clearly justified and proportionate in meeting the stated purpose and be suitably validated. It should always involve human intervention before decisions are taken which affect an individual adversely”.
The last sentence is key for the deployment of all super recognisers where autonomous facial recognition systems are used. At the moment, the Code is only enforceable for publicly owned CCTV systems – but we support this being extended, whether voluntarily or by statute, into the private sector.
`Dr Davis has been fundamental in generating research into the Super Recogniser discipline and, indeed, the online testing regime, which has now been taken by over 4 million people around the globe. Josh has kindly agreed to be the Honorary Chair of the Associations Validation Panel, we are hugely grateful for his help and assistance and long may Greenwich University leads the world in this area. One of the tests used for validation has been taken by over 5,000,000 people although more importantly, all validation tests have be trialled by 100s of super recognisers.
During the preceding three months those subscribing to the Association of Super Recognisers – ASR. has more than doubled. This has provided a huge boost to the organisation.
The second ‘licentiate’ training course took place in September and I am delighted to inform you that the pass rate was 100%, many congratulations to all those who attended
An ‘Admissions Ceremony’ will take during in 2019.
The Biometrics Institute held a Facial Recognition workshop, which proved invaluable. Led by Patrick Growther from The National Institute of Standards and Technology – NIST who delivered and excellent presentation ‘Changing the Face of Facial Recognition’ he rightly described the facial recognition industry as ‘In the midst of a revolution,’ both technological and human with the greater understanding and deployment of super recognisers.
He also talked the growing issue of ‘morphed images or blended images.’ Morphed images is when 2 or more, usually 4 in order to present a number of different people on the same image. This is used to bypass Artificial Intelligence Facial Recognition systems. Read more on WWW.NIST.gov and search for morphed images.
The article in the Independent newspaper , 12th September with the headline ‘Super-recognisers employed by UK police to hunt criminals may not be identifying people correctly, study suggests. www.independent.co.uk/news/science/facial-recognition-super-recognisers-police-met-salisbury-attack-british-science-festival-a8533361.html
This article rightly highlights that not all super recognisers who are deployed are able to remember and identify criminals. Research conducted at Bournemouth University has revealed inconsistencies in super recogniser abilities. This is not news to the ASR and why it has been established.
The article refers to a new research paper www.cognitiveresearchjournal.springeropen.com/articles/10.1186/s41235-018-0116-5 The conclusions of which are that current screening protocols for super recognition need to be expanded. The super recognisers used in the study were self identified and had not undergone the rigorous training course that Licentiates of the ASR are required to complete.
Engagement with a number of commercial organisations has been requested with the view of forming strategic partnerships. Discussions are well underway – watch this space.
Thanks to Kenny Long from Super Recognisers International for his feedback from some of the positive feedback received from licentiate ASR professional super recognisers.
The first criminal to be brought to justice by a justice by a non-police Super Recogniser.
Mike commented "This shows the value of non-police Super Recognisers working in the security industry. The police have now linked the suspect to several more offences - so his crimes will now be taken seriously."
On 7 Sep 18, an American lady had her handbag stolen from a wine bar in the Charing Cross area. Ex-cop and now Super Recogniser International Ltd, Chief Operating Officer, Kenny Long, was shown this image:
One week later on 13/09/18, Kenny was in London in the Charring Cross and as he was walking along he took the following picture of an Anti-Donald Trump March.
Moments after taking this photo Kenny saw a male who he believed to have been the suspect who had committed the theft at the Wine Bar the week before.
He followed the male and knowing that the Anti-Trump march was going on and there would be plenty of police officers on duty.
The male was getting very close to other members of the public and appeared to be looking in their bags.
As a result he relayed the information to police officers and they stopped the male. After explaining what evidence he had, the officers arrested the suspect.
· Picture below when spotted and arrested.
· The suspect (Hareb) pleaded guilty to theft at Westminster Magistrates Court on 10/10/18.
· He will be sentenced on 31/10/18 at Westminster Magistrates Court.
All in all, a great result for SR’s demonstrating our skills
Mug Shot Memories
Before anyone asks, no, this is not Gilly having a bad hair day, even non-Super Recognisers can probably recognise a lot more faces than they think possible – normally around 5,000.
“It seems to be overkill” says Rob Jenkins at the University of York in the UK. That is because humans lived in groups of around 150 people for most of their evolutionary history, so the usual idea is that it makes sense for us to recall about this number of faces.
To test this idea, Jenkins and his team asked 25 people to spend an hour writing down the people they knew personally for whom they could form a clear mental image of their face.
On average, participants listed 40 people in the first 5 minutes of the exercise and 21 in the final 5 minutes. From these rates, the team calculated that a typical participant would have listed 549 people if they were given unlimited time.
Then they tested recognition of famous faces by showing images of 3441 public figures from the likes of film, business, politics and sport. Each participant saw from the likes of film, business, politics and sport. Each participant saw a different picture of each figure on different days. If they said they recognised the celebrity in both pictures, then that person was considered part of their ‘facial vocabulary’.
Participants recognised about 30%of these faces. Based on this, Jenkins and his team estimated that most people recall about 5,000 faces.
(From The New Scientist Magazine - Number 3200 - Proceedings of the Royal Society B.doi.org/cvqb).
Helping to Counter the Terrorist Threat Using Face Recognition
Helping to Counter the Terrorist Threat using Face Recognition: Forensic Media Analysis Integrated with Live Surveillance Matching – an article from Allevate
Against the backdrop of budget constraints, threats from terrorism, organised crime and public disorder continue to rise. Authorities can remain resilient through the targeted application of technology. Advances in face recognition coupled with the mass availability of digital media and continuously cheaper computing provides unique opportunities to enhance the efficiency of forensic investigations to enhance public safety. Processing of digital media can be automated in a virtualised and elastic computing environment to identify and extract actionable intelligence. Processing is scalable, continuous, consistent and predictable. Analysts can focus on investigating and confirming suggested results rather than watching countless hours of media in the hope of stumbling across intelligence. Such a centralised platform can also be used to search in near real-time faces from any number of remote cameras against centralised watchlists of individuals of interest.
1 A Need for Enhanced Safety and Operational Efficiency
Risks are increasing. Recent events demonstrate that the threat landscape is substantial and becoming more fragmented, consisting of a greater number of smaller and less sophisticated plots. The targeted application of technology can play a key role in improving the efficiency of our police and intelligence agencies and maintaining readiness to both disrupt and respond to major events.
2 A Relentless Increase in Digital Media
The increase in media is relentless. Law enforcement and intelligence agencies have amassed large collections of video and photographic information from multiple sources such as:
Digital Forensics (confiscated phones, computers, flash drives etc).
Open Source Intelligence (Internet and Dark Web).
Crowd- sourced from members of the public (HD cameras on mobile phones are ubiquitous).
Police Body Worn Video. CCTV.
When tragic events or social disorder occur, investigators have a long and arduous task of reviewing countless hours of media, generally with a varying degree of concentration and scrutiny.
A solution that minimises manual effort in the extraction of actionable intelligence from amassed media by automating this process with a consistent and repeatable level of scrutiny will deliver concise and consistent information in a fraction of the time taken by operators undertaking the task manually.
3 An Automated Media Processing and Exploitation Solution
Police, intelligence and other public order agencies can benefit from the application of a powerful media processing solution designed to ingest, analyse and index, in an automated fashion, very large quantities of media from multiple sources to transform them into usable assets. Utilising virtualised and elastic computing environments enables the platform to be rapidly scaled up and down in response to unfolding events.
Once processed, agencies can analyse and make use of the extracted assets and manage them in a centralised repository of information. Data links, associations and metadata inferences can be managed across the whole dataset by multiple users from a single common user interface. Backend processing services are run in a cloud computing environment, the capacity of which can be configured and incrementally scaled up and down to meet an organisation’s changing demands; peaks arising from specific events can be easily accommodated.
Automatically find, extract and index faces to enable biometric and biographic searching of media.
Create and manage watchlists of people of interest.
Find and cross-reference all media instances in which a person of interest has been seen.
Identify, locate, and track persons of interest, their associates and their activities across all media.
Discover, document and view links between people of interest, their activities and networks.
Use of metadata (including geo data) to enhance investigations and association of data.
Integration into existing system environments, databases and components.
3.1 Incorporating Other Detection Capabilities
In addition to face recognition, other detection engines can be incorporated, such as:
Biographic filtering and Fuzzy Match capability.
Automatic Number Plate Recognition. (ANPR)
Object / Logo Recognition.
Vendor independence allows the use best-of-breed algorithms. Newer and better algorithms (COTS and GOTS) can be plugged in without having to replace the entire platform.
3.2 Working with Geo-Location Data
An increasing amount of media is captured on devices affixed with location determining technology. Often, this geo-location data is incorporated into the media metadata, thereby providing potential to further enhance the analysis of media. Geo-location can be used to:
Compartmentalise and refine analysis by location of media creation. Overlay location of proposed matches onto maps. Chart movements of individuals of interest by location and time of sightings. Link individuals at the same location and time even if they do not appear together in media.
3.3 Architecture and Integration with Existing Systems
In addition to utilising COTS components, open standards and cloud-computing architecture to enable massive scalability, a well delineated scope of functionality and open API enables:
Flexibility in customisation and integration with existing systems and workflows. Well-defined mechanisms of loading data and automating ingestion of media. Dynamic alteration and sharing of watchlists, media, system-generated results and operator analysis.
3.4 Hosting, Cloud and Virtualisation Options
Full architectural flexibility enables flexibility of hosting options. Organisations can elect to:
Take advantage of IaaS and SaaS options on public sector hosting offerings. Fully self-host the solution on private and secure premises and datacentres. Deploy in a hybrid manner.
Indeed, managed AWS or Azure offerings can be utilised to bulk process media, utilising nonreturn gateways to propagate identified sensitive data to more secure facilities.
3.5 Working Hand-in-Glove with Trained Forensic Investigators
Humans will always remain the critical and essential part of intelligence analysis; such solutions do not replace the intricate skills and knowledge of trained investigators. Rather, the operator is enabled to intelligently direct and apply their training at suggested results, eliminating the necessity of rote viewing of countless hours of media either in a sequential our random fashion.
Integration of enhanced verification, charting and mapping tools enables operators to conduct detailed analysis of suggested matches and identifications.
4 Potential Use Cases
There are multiple applications of a solution as described herein within military, law enforcement, intelligence and public-site security agencies. These are summarised into four broad categories:
4.1 Time Critical Investigations, Media of Critical Importance
Often, authorities need to quickly process evidence to identify and apprehend individuals. The scale of the investigation can be huge and the amount of media that needs to be processed massive.
The media acquired in these instances can be of such critical importance that the authorities may choose to review it all in its entirety. However, immediate and decisive action is critical. Rather than sifting through the media in a random or sequential fashion, a media analysis solution can quickly direct the investigators to portions of the media that are most likely to deliver immediate results. Full review of the media can be conducted afterwards.
4.2 Bulk Ingestion of Media Arising from Criminal Investigations
During routine operations or investigations, authorities may recover significant quantities of media from multiple sources that need to be processed to further the investigation or to assist in building an evidence base for prosecution. Examples include:
Military or counter-terror officers raiding terrorist facilities. Specialist organised crime investigators raiding organised crime offices. Child protection officers raiding premises of individuals or organisations involved in child exploitation.
Automating processing provides investigating officers an overall summary of the contents including focus areas for further investigation.
4.3 Continuous Background Processing of Media Sources
Authorities may as a matter of routine have access to masses of media which may contain actionable intelligence, but typically would never be viewed or processed due to a lack of resource. Intelligence in this media may be missed entirely and never acted upon.
This media can now be bulk ingested and processed in an automated fashion to flag relevant intelligence, using operator-controlled criteria, to the authorities as required for follow-up processing.
Routine and automated processing of accessible media can flag actionable intelligence that may help disrupt future attacks.
4.4 Near Real-Time Watchlist Checking from Live Surveillance Cameras
By integrating any number of remote surveillance cameras to such a centralised matching platform eliminates the need to install and maintain costly local software and hardware to perform local face matching as well as the need to store potentially secure watchlist data locally at the camera locations. The problems associated with live streaming of HD media over low bandwidth network connections is resolved through the application of local face-detection and cropping; only small image files of cropped faces need be sent to the central data centre over encrypted channels.
4.4.1 Centralised Archive of “Seen Faces”
In addition to submitting search probes to the server for searching against one or more watchlists, search probes can be enrolled in a “seen faces” archive which can be interactively or automatically searched (using face recognition) by investigators or when submitting videos for processing.
5. A Compelling Business Case
The solution can be made available using a compelling SaaS model. The open and standard nature of the solution ensures it can run in existing on-premise datacentres or outsourced to secure hosting partners.
Whilst the human operator is an essential part of intelligence analysis, an entry-level system empowers the analyst to process up to an order-of-magnitude more media on a daily basis. This enables trained operators to apply their expertise in a more focused manner than manually watching hour upon hour of media.
Efficiency is dramatically boosted by bulk processing media 24x7 at a constant and predictable level of focus and accuracy: operational staff can focus on analysing results.
Book Review – By Brig Jeff Little OBE
Violet Hill - Written by Henrietta McKervey
Although you may have seen this book on the ASR website, Jeff Little has kindly conducted an more in depth review – a fascinating read!
This is a daring, challenging and emotional book to read. I found parts of it difficult to push through at times but the finale made it a very worthwhile journey. I finished the concluding chapter late one night with a smile on my face, then another smile, then a tear in my eye and lastly a decision to make as to which version of the life of the mysterious Phyllida Hartigan and her dreadfully scarred face I wanted to believe. Never have I read a book and been left with so many mixed emotions or choices to make.
Two parallel investigations form the central theme. The first takes place in 1919 at the end of the First World War with the country trying to come to terms with the shockingly injured and distressed soldiers now reappearing on the streets, with Will Dockery wearing a tin mask to protect his dreadfully disfigured face and his brother Davey seeking to win the affections of Chrissie Hillman who works for Violet Hill. Violet Hill Investigations, the only female private detective agency in London, faces financial ruin until it becomes involved in the strangely spiritual world of Sir Arthur Conan Doyle himself, creator of Sherlock Holmes no less, who like many others, is seeking to contact his son who was killed in a recent battle. The mysterious J C Selbarre and his assistant Mademoiselle Du Ferge are the centre of the case involving the London Spiritualist Association. But is Selbarre a fraud who is exploiting the dead and the mourners of war or is he really and truly a path to the lost souls of a whole generation of young men killed in bloody action?
The other investigation takes place in 2017 with Detective Constable Susanna Tenant, a super recogniser in the Metropolitan Police on the hunt for a dangerous serial attacker. Just as she recognises the perpetrator and is about to arrest him, Susanna agonisingly loses her SR skills, in a rather bizarre and difficult to believe accident – although as she says herself, many ‘bizarre and difficult to believe’ crimes happen every day out on the streets. While she is convalescing from the accident with a broken wrist and a sore head, she tries to switch back on her super recognition powers, she befriends a strange serial shoplifter who adds a most human, emotional and moving element to the tale. Gareth Slatten, a footman and Henry Gardiner, who is remembered by a blue plaque in Susanna’s home, provide the bridge back to the 1919 investigations.
The links in the chain of time between the 2 cases provide much food for thought.Violet and Susanna both face danger, ruination and death in very different times and very different circumstances - but in the end, they both use their detective skills to find the vital evidence to prove their cases and expose the dreadful truth – in Susanna’s case with shockingly violent consequences and Violet facing a most untimely death leaving the reader truly shocked and dismayed.Super Recognisers will love the book – it is well worth persevering with for the ending is divinely painful.
Many thanks to all those who contributed to the newsletter especially Brigadier Jeff Little OBE MBA CGIA FICPEM FSyI for his substantial contribution.
Do you have an article or story to tell and would like to have it published?
0800 773 4869
Chief Executive Officer