• Phone: +1 844-COGNITRO
  • info@cognitro.com
Stay Connected:


A Fast Face Recognition Model

 Biometric face recognition technology has received significant attention in the past several years due to its potential for a wide variety of applications across sectors. Compared with fingerprint/palmprint and iris, face recognition has distinct advantages because of its non-contact process. Face images can be captured from a distance without touching the person being identified, and the identification does not require interacting with the person. In addition, face recognition serves the crime deterrent purpose because face images that have been recorded and archived can later help identify a person. Cognitro has successfully developed a face recognition algorithm,  FastMatch.Ai for authenticating faces at record speeds. The model has been deployed on large scale at several governments facilities and private organizations.

Deployment Options

FastMatch.Ai On-Premise
With just 2-3 face images per person captured with regular web-cams, we can train a model to recognize any number of subjects and deploy on inexpensive CPU-based machines.
FastMatch.Ai on Cloud
Eliminate the burden of hardware cost and setup by deploying face recognition As-A-Service where authentication is then done through an API or web-interface or mobile app.
FastMatch.Ai in the Box
Take it to the edge by embedding the recognition algorithm on a GPU-based AI chips so you can achieve decentralized, mobile and wireless applications such as drones and with the highest accuracy.

Major Benefits


The ability to scale up the model for up to 70M faces of subjects in  the database with no-loss in performance


Augment the existing model by seamlessly adding new images on-demand
without training


Train or retrain the FR model dynamically with minimal intervention and with no expertise needed


Unsurpassed face recognition speed of
300 milliseconds per face on a  CPU-Based environment


Minimal investment on hardware, optimal for CPU–based, comparable with GPU-based computing


Interface with FR model using a dynamic REST-API that can be integrated with external systems


Identity Name Matching

 Names are important elements in government transaction, banking compliance, identity protection and verification. However, with so many similarities across different names and languages, spellings, aliases or nicknames, order of first, initial and last names, many situations can arise such as record duplicates, identity fraud which may pose significant risk levels. Cognitro has worked with a number of organizations to alleviate those types of risks by leveraging the latest in AI and deep learning to match any given name against a database of other names.

Example: 1) Moammar Khadafy  ⇔ 2) Muammar Gaddafi 

WhosWho Features
We tested the model accuracy and performance against other famous transliteration systems such as Rosette System using standard text analytics metrics such Levenshtein distance and Word Error Rate. Our system was shown to have surpassed most well-known system by an accuracy margin of 7%-8%.