Face Recognition in Finding People
Development in Face Recognition can be designated as “Fast & Furious” with improved performance and broader acceptance in different application over the last decade. Comes with this is the question about the violation of privacy by the user of face recognition. I will pen my thoughts on this in a separate article and focus only on the generic use of finding people using face recognition.
The Law Enforcement Agency (LEA) is the traditional user of face recognition. In recent years, we saw the application of face recognition in payment, customer service, access control, and time attendance. While finding people remains the same, the application differs depending on the type of user.
Face recognition can be broadly classified as either 1:1 or 1:N. In 1:1, an individual’s face is used to compare against who this individual claims to be against the database’s stored record. In 1:N, an individual’s face is used to search against the database to find the closest match. In the 1:N scenario, N’s value could be 10s, 100s, 1,000s, or a few million records.
The ability to find people is crucial for using face recognition in LEA and the most common application is in immigration, city surveillance or when verifying someone’s identity. For LEA, the techniques are often more comprehensive and complex, not limiting to face recognition. I have written an article “Layered City Surveillance” some time back on this matter.
Apart from considering technology adaptation, human factors are another consideration that affects such a project’s outcome. It directly affects the effectiveness of LEA in finding people or performing their investigative work. Human in Human Factors refers to both people who use the system and people LEA is trying to find.
When running the first face recognition project for Singapore’s homeland ministry more than ten years ago, we focus solely on the technology aspect. For every ground challenge, we seek an answer from the technology aspect, which is not a holistic approach. Fortunately, the attention generated by face recognition in recent years increases human factors’ awareness among the face recognition system’s implementor.
In any industry, there’s likely to be a counterforce developed to render the use of specific technologies or techniques ineffective.
- Electronic Counter Measures — Electronic Counter Countermeasures
- Cyber Attack — Cyber Defense
- Unstoppable Force Paradox, which is also Shield & Spear Paradox
I was an Electronic Warfare Specialist when servicing my regular service in the Republic of Singapore Navy. ESM, ECM, ECCM was what I am dealing with constantly for my work.
Thus, it’s natural to have anti-face recognition to defeat its effectiveness when face recognition is gaining popularity. On the same note, there will also be anti-anti-face recognition techniques, and it just keeps going. There are a lot of methods floating online that said to render the use of face recognition ineffective. It ranges from Make-up, Markings on the Face, Anti-CCTV Glasses. I remember seeing a transparent glass technology that aims to spoof face recognition in recognising an individual’s correct identity.
Some of the face recognition’s key strength includes non-intrusive and the ability to have a long standoff range. In theory, as long as the image sensor has a clear sight of the person’s face, face recognition can use that image to perform matching against the database. The question is, why are we so focus solely on the face when there are other biometric and non-biometric signatures that an individual is exhibiting and can be taken non-intrusively?
- Clothing, bag, headgear, body markings, etc., can be unique to a group that an individual is associated with.
- Each of us walks in a specific manner or pattern. To search for an individual gait can also be used. Gait Abnormality can be derived and learn to detect abnormal, which potentially include neurologic disorder, state of duress, etc.
- The mobile device is now our extended identity, and we basically cannot travel without them. These devices emit a specific signature that LEA can use to identify an individual.
The development of anti-facial recognition techniques and method support such a trend from increased awareness of such technology and evasive techniques makes finding people using face recognition more difficult. While face recognition technology will find its way to counter this, the other camp will also fight back.
I am leveraging on a famous quote by IRA after failing to murder Margaret Thatcher in the Brighton Bomb.
“Today, we were unlucky, but remember we only have to be lucky once — you will have to be lucky always.”
We can now turn the table around using this analogy. While someone with ill-intent might conceal one’s identity by hiding face and doing all sorts of concealment, they need to hide all that they can from sight. While they need to hide everything, a holistic platform need to detect only one signature to increase the probability of getting that person.
There’s more than a way to find an individual apart from finding their face.
