«Expasoft» technologies based on Machine Learning and Big Bata analysis.
Our expertise in the field of Big Data analysis and Machine Learning allowed us to create a unique of its kind technology for processing, analysis, storage and management of large data.
Thanks to the technologies of big data analysis generated in our work, we can offer both development of ready solutions for various industries and businesses, as well as integration of these technologies to solve problems in the field of data processing for IT companies working with b2b and b2c customer service.
We prepared a set of intelligent services, each with its own API.
So let us introduce the functionality and objectives that our technologies address:
1. Face recognition
It includes two main practical cases:
- Formation of reference images database (will be used for search of the original);
- Demonstration of the recognition algorithm;
In both cases the user can upload one or more predefined images or take a photo with a webcam.
In order to test the image recognition service, you need:
- to follow the link to the relevant section of the site, presented below;
- Select "Detection and Face Recognition";
- Sign up and upload a reference face image over which you will identify other images;
- Identify the images of faces by uploading them into the system.
In practice systems of face recognition solve two main problems:
- Verification problem;
- Identification problem;
Verification problem is to compare two photographs in order to understand whether the people are the same person or different people (one-to-one). This problem often occurs in security systems when it is necessary to understand whether the person is who they claim to be.
Identification problem is to identify a person in the photo from a predetermined set of persons (one-to-many). The problem of identification is technically more complex as it is a problem of multi-class classification.
Face recognition technology finds its practical application in retail, service providers, identifying individuals in companies and other organizations that are faced with a large flow of people on a daily basis.
An example of implementation of this technology is "Score Face" recognition service designed by "Expasoft".
2. Identification of speakers
The service implies 2 main practical cases:
- Formation of the reference voice recordings database;
- Demonstration of recognition quality;
In both cases the user can upload one or more prerecorded audio files, or record them from a microphone of a smartphone or computer.
Identification of speakers service is used in various fields of business and industry for authentication and voice recognition of employees, customers and visitors.
3. Biometric authentication
Company "Expasoft" developed an authentication service that implements the following scenario:
- The user passing authentication is sent a text message;
- The user reads the message on the microphone and sends it to the service;
- the voice recording is checked to match the message + voice sample is checked with the voice DB;
Thus, a three-factor authentication is performed:
- availability of a registered phone;
- matching the voice with a reference sample;
- matching the contents of the voice message with the text sent;
The service uses an original technology, which allowed:
- to reduce the size of voice sample to 5 sec;
- to superceed a similar solution from Microsoft on a standard set of data of MOBIO Database.
Demo-version of the service allows to perform the following tasks:
- basing a person;
- person authentication;
To test the service the user needs to upload data about the person to the database. To do this, you'll need to:
- Turn on the web camera,
- look into the camera ,
- read the automatically generated text,
- upload the result.
Authentication is performed by user reading the system generated messages to the camera, then the system analyzes the results and gives a message of authentication degree.
Biometric authentication technology is practically applied in various fields and business areas in which there's a problem of security and privacy.
4. Documents recognition
Users need to upload one or more predefined images.
For each image, the system provides a response on the recognition of a certain class of documents.
Recognized classes of documents:
- Russian Federation internal passport;
- Russian Federation international passport;
- State pension insurance certificate;
- Driver's license;
- Vehicle registration certificate.
Document recognition technology has been successfully applied for solving a variety of tasks in business and industrial sectors.
5. Recognition of sound events
"Expasoft" company and "NSU-Expasoft" Data Flow Analysis Lab implemented the technology of audio events recognition.
With a high degree of certainty (at least 98%) we can recognize:
- baby crying
- dog's bark
- sound of breaking glass
- fire alarm
To test the service please use the beta-version, link below.
Upload one or more prerecorded audio files, or record from smartphone or PC microphone and click the "Recognize" button.
To see the services and how they work, visit our website: