Human Recognition Module
Last updated
Last updated
The first step to establishing a network of verified human beings is a human biometrics recognition system that is capable of verifying the uniqueness of human beings with accuracy, reliability, inclusiveness (non-invasive and easily accessible hardware), and spoof-resistance.
Challenges of Verifying Human Uniqueness
At first glance, verifying human uniqueness with biometrics may seem straightforward since we have already witnessed the mass adoption of biometrics-based authentication systems in the past decade (e.g. Apple’s TouchID and FaceID).
However, upon further thought, the accuracy requirement of human uniqueness verification is significantly higher than biometric-based authentication — uniqueness checks require not just verifying one submitted biometrics sample be similar to the profile under question (”1-to-1 matching”), but also that the submitted sample be dissimilar to all other profiles in the profile universe (”1-to-N” matching, with N potentially in the order of millions, if not billions). This level of accuracy is, sadly, not yet seen in currently available mass-adopted biometrics systems.
We also face another conflicting challenge on the privacy front — any biometrics system that is “too accurate” also poses significant privacy risks to the users, especially if the collected biometrics data are saved and managed online. For example, collecting a user’s own and family-tree DNA profiles (!) would satisfy the accuracy requirement to verify human uniqueness, but the average user would feel uncomfortable with the privacy risks posed by this excessively invasive technique.
The Humanity Protocol Approach - “Right Amount of Biometrics” with Palm Recognition
In designing Humanity Protocol’s human recognition module, we tackle both the technological and privacy challenges head-on and find the optimal trade-off between the conflicting goals. The solution is a palm recognition technology with “the right amount” of biometrics signature — sufficiently accurate for the 1-to-N matching problem and respectful of user privacy to the maximum extent while still being reliable, inclusive and spoof-resistant.
Similar to the rest of the modular architecture, the human recognition module will be developed in 2 phases:
Phase 1 utilizes a palm print recognition software program that can be installed on users’ smartphones, ensuring a low barrier of entry for individual users. This method allows for the generation of an RGB image using a smartphone camera in natural light, leveraging the mature technology of palm print analysis.
Phase 2 will introduce palm vein recognition, employing a specialized (but still low-cost and easily accessible) device with an infrared camera that can be physically connected to a smartphone. This stage redirects users to the Humanity App, where the unique vein patterns in the palm are analyzed for even more precise identity verification — a study by Fujitsu (2006) used 140,000 palm profiles of 70,000 individuals, palm vein scanning demonstrated a false acceptance rate of less than 0.00008% and a false rejection rate of 0.01%.
In both phases, captured images are processed and enhanced so that principal palm features can be extracted by Humanity Protocol's AI model (based on convoluted neural networks CNN), further elevating the speed of the process (<0.1 seconds) and level of robustness (e.g. to measurement errors and varying lighting conditions). When both palm print and palm vein are used in combination, Proof of Humanity is accurate enough to ensure that there is a unique biometric signature covering the entire human population.
Our approach pushes the efficient frontier in all the dimensions below:
Accuracy
The human palm print has a large surface area with a complex set of features (skin lines, loops and creases) that is truly unique to each individual, carrying more information about an individual than other biometrics, such as fingerprints and iris scans. Palm vein recognition technology goes further by analyzing how the haemoglobin in our blood interacts with infrared light and capturing images of the intricate network of veins within our palms.
Reliability
An individual's palm features, especially vein patterns, remain stable throughout one’s lifetime. This allows for a stable signature for HP users and reduces the need for repeated updates or re-enrollment.
Robustness to Spoofing
Our proprietary technology captures palm features in two spectral bands: palm print via visible light and palm vein via infrared light. This combined approach makes spoofing by non-human agents nearly impossible. We will also enforce production policies (e.g. a limited number of trials per device per user) that will further snuff out any opportunity for malicious behavior.
Inclusive
Our Phase 1 palm print scan is designed to run on easily available hardware, such as your cell phone camera. The Phase 2 palm vein DePIN device, while more specialized, is still easily affordable and accessible by individual users. In both phases, the entire process is contactless: simply hover your hand over the camera.
No mess, extremely fast, and no expensive hardware required — designed for everyone to participate.