Could a face-reading AI tell police when suspects are lying?
Could a face-reading AI ‘lie detector’ tell police when suspects aren’t telling the truth? UK start up is in talks with Indian and British police for trials
- Micro-expressions are subtle facial movements that can betray hidden emotions
- They are impossible to suppress and could be used to tell when someone is lying
- London-based Facesoft has been training an AI to spot these micro-expressions
- The firm has been liaising with police in the UK and India about practical uses
Forget the old ‘good cop, bad cop’ routine — soon police may be turning to artificial intelligence systems that can reveal a suspect’s true emotions during interrogations.
The face-scanning technology would rely on micro-expressions, tiny involuntary facial movements that betray true feelings and even reveal when people are lying.
London-based startup Facesoft has been training an AI on micro-expressions seen on the faces of real-life people, as well as in a database of 300 million expressions.
The firm has been in discussion with both UK and Mumbai police forces about potential practical applications for the AI technology.
The latter are reportedly interested in using the technology as part of crowd control measures, with the algorithm detecting when an angry mob might be forming.
Scroll down for video
Forget the old ‘good cop, bad cop’ routine — soon police may be turning to artificial intelligence systems that can reveal a suspect’s true emotions (stock image)
The emotion-identifying tech has been developed by British firm Facesoft, who describes micro-expressions as a form of so-called ’emotional leakage’.
Micro-expressions were first recognised as an indicator of deception by 1960s psychologists, who saw the facial movements on suicidal patients who were trying to cover up their strong negative feelings.
‘If someone smiles insincerely, their mouth may smile, but the smile doesn’t reach their eyes — micro-expressions are more subtle than that and quicker,’ Allan Ponniah, Facesoft co-founder and consultant plastic surgeon, told the Times.
Dr Ponniah, who practices at the Royal Free Hospital in northwest London, developed an interest in AI through its potential to reconstruct patient faces.
‘[Micro-expressions] occur on various regions of the face, last only a fraction of a second and are universal across cultures,’ wrote experts from Carnegie Mellon University, Pittsburgh, who are not part of Facesoft but also study micro-expressions.
Dr Ponniah and his colleague Stefanos Zafeiriou — an AI expert from Imperial College London and Facesoft’s chief technology officer — said that the system would be able to flag key sections from hours of suspect interviews.
These parts could then be reviewed in detail by police psychologists, they added.
Facesoft have trained their algorithm using a database of 300 million high-res images of human faces, covering all age groups, ethnicities and genders.
Facesoft has labelled each image with one or more particular emotions that they display — including happiness, fear and surprise — alongside recording the intensity and power of the corresponding feelings.
From this dataset, the researchers can create billions of synthetic faces to help teach their AI to recognise more people more accurately.
The use of so many artificially generated faces allows for more accurate identification of the algorithm’s subjects and helps avoid the ethnic and gender biases that can arise from the use of a limited-scale training dataset.
The firm has been in discussion with both UK and Mumbai police forces about potential practical applications for the AI technology
To teach their system how to pick up on micro-expressions, Facesoft researchers have also been using a high-speed camera to record volunteers — who had been instructed to suppress all facial movements — as they watched films.
Researchers then analysed the footage, identifying movements where the participants ‘true emotions’ leaked out.
After tagging these micro-expression leaks, the team fed the data into their AI, whose design, modelled roughly on the structure of the human brain, is referred to as a deep neural network.
In a recent US government test, Facesoft’s facial recognition software was ranked better than both its American, European and Russian competitors when handling so-called ‘wild’ images that only partially show the subject, such as taken by CCTV.
Wild images are more technically challenging to match in comparison with, say, passport-style, face-on portraits.
‘We’re delighted to be ranked so highly by the Facial Recognition Vendor Test,’ Dr Ponniah told MailOnline.
‘We’re looking forward to making further improvements that will only increase our accuracy measurements,’ he added.
‘Facial recognition has the potential to be a great asset to society and public safety and the developments we’re making mean the benefits will only expand.’
The researchers believe that Facesoft could eventually be used to identify when a crowd of people might be able to evolve into an angry mob (stock image)
The researchers believe that their algorithm could eventually be used to identify when a crowd of people might be able to evolve into an angry mob.
The tech has firm has already been in discussion with police officials in Mumbai, India about developing a system based on this particular monitoring concept.
Facesoft has also been in discussions with police forces in the UK about possible applications for their facial recognition technology.
Not everyone is enthused about the potential incorporation of AI into police work, however.
In April 2019, the Partnership on AI — whose members include Amazon, Apple, Facebook, Google, IBM, Microsoft and academics — came out against AI algorithms designed to help law enforcement in sentencing, bail, parole or probation decisions.
Although such systems are growing in popularity in the US, the partnership expressed concern that these algorithms have the potential to be biased, opaque or simply may not function as intended.
Logan Koepke, senior policy analyst at Partnership on AI member Upturn, told Bloomberg that the group’s report ‘highlights, at a statistical and technical level, just how far we are from being ready to deploy these tools responsibly.’
HOW DOES FACIAL RECOGNITION TECHNOLOGY WORK?
Facial recognition software works by matching real time images to a previous photograph of a person.
Each face has approximately 80 unique nodal points across the eyes, nose, cheeky and mouth which distinguish one person from another.
A digital video camera measures the distance between various points on the human face, such as the width of the nose, depth of the eye sockets, distance between the eyes and shape of the jawline.
A different smart surveillance system (pictured) can scan 2 billion faces within seconds has been revealed in China. The system connects to millions of CCTV cameras and uses artificial intelligence to pick out targets. The military is working on applying a similar version of this with AI to track people across the country
This produces a unique numerical code that can then be linked with a matching code gleaned from a previous photograph.
A facial recognition system used by officials in China connects to millions of CCTV cameras and uses artificial intelligence to pick out targets.
Experts believe that facial recognition technology will soon overtake fingerprint technology as the most effective way to identify people.
Source: Read Full Article