Our process involves looking for certain people
which are carefully selected based on location and time.
How Matches Work
Algorithm measures facial features
The system can search through non perfect images however the following must be visible:
- EYES
- NOSE
- SHAPE OF THE FACE
Images with multiple persons
The system has the capability to identify multiple faces within an image, e.g. if you have a photograph of a crowd at a football match or a nightclub still the system can identify more than one person.
The more of the face that is seen the better the result
The whole system is based on an Algorithm which measures a number of facial features and their distance from the eyes, therefore the more head on the image the better the result.
The system is reliant on combining cutting edge technology with an experienced operator
We are able to achieve a high success rate by identifying operators that are proficient at recognising faces which compliment the technology.
Data Management & Retention
We are acutely aware of the need to balance privacy and security when utilising and deploying AFR Locate.
WATCHLIST IMAGES, METADATA AND BIOMETRIC TEMPLATE
DELETED IMMEDIATELY POST DEPLOYMENT IN ANY CASE WITHIN 24 HOURS
AFR LOCATE DEPLOYMENT WATCHLISTS
UNIQUE FOR EACH DEPLOYMENT, WHEN THE DEPLOYMENT IS OVER MORE THAN ONE DAY THEY ARE UPDATED EVERY 24 HOURS.
NON ALERT IMAGE TO INCLUDE BIOMETRIC TEMPLATE
NOT RETAINED WITHIN THE APPLICATION
IMAGES INCLUDED IN MATCH REPORT
DELETED IMMEDIATELY POST DEPLOYMENT IN ANY CASE WITHIN 24 HOURS
MEMORY STICK USED TO TRANSFER WATCHLIST FROM SHARED FOLDER TO AFR APPLICATION
DELETED IMMEDIATELY AFTER WATCHLIST TRANSFER
CCTV FEED (NO AUDIO)
31 DAYS
MATCH REPORT
31 DAYS
AFR LOCATE OPERATOR LOG
31 DAYS
Equipment
All solutions can be used both indoors & outside.
MOBILE AFR VEHICLE
Used for Public Events or Rapid Deployments
PORTABLE PTZ CAMERA
Pre-Planned where distance or 360 options are required.
PORTABLE STATIC CAMERA
Pre-Planned, ideal for narrow pinch points or building.
FAQ's
A statement from South Wales Police regarding the Judicial Review outcome can be found in the Blog section of the website. Alternatively an FAQ regarding AFR and the judicial review can be found here: Judicial Review Appeal FAQ
Typical deployments would obtain images from South Wales Police and Gwent shared custody database.
If the system does not make a match your image is not retained.
All AFR Locate deployments are overt and prior to each deployment we will use social media to advertise the deployment, during the deployment signage will be used. All AFR vehicles are clearly marked.
The image will be deleted immediately post deployment and in any case within 24 hours.
No, avoidance of the AFR cameras in isolation does not in itself constitute grounds for search or arrest.
A police officer may want to speak with you to confirm your identity.
The officer will explain why you have been spoken to and where possible invite you to see the technology. You will also be provided a Privacy Notice (see resource) which will provide further information.
Having reviewed 18 months of deployments there is no evidence to support this claim (see resource).
Watchlists for use with Live Facial Recognition (AFR Locate) must be specific to an operation or to a defined policing objective. Must be based on the intelligence case, reviewed before each Deployment to ensure that all images meet the necessity and proportionality criteria for inclusion, and the make-up of the Watchlist should not be excessive for the purpose of the LFR Deployment. Watchlists are primarily made up of individuals suspected of committing a criminal offence, wanted on warrant, unlawfully at large or there are bail conditions or court order or other restriction that would be breached if they were at the location at the time of the Deployment. They can also include missing persons who have been deemed increased risk or include individuals to prevent a risk of harm to themselves or others. This term will be informed by the intelligence case, using LFR to facilitate their location is necessary to manage the risk of harm identified