Amazon Rekognition validación de identidades en evaluaciones online

Case StudyCaso de EstudioCas d'Estudi

A prestigious online university implements an identity validation system for online assessments

We fulfill the need of an important online university to develop a tool to validate identities in online assessments. In 6 days, the identity and authorship of the students was verified in 70,000 online exams taken with a customized system.

The system developed for the university is based on Artificial Intelligence and image recognition services (Amazon Rekognition) of the AWS cloud, allowing the assessment to be completely online, without the need for any face-to-face act.

Challenge

The pandemic situation forced us to move from on-site exams to a 100% remote scenario, with online assessment processes that must guarantee that the person taking the test was the right one.

Solution

Our solution was to develop a customized identity validation system based on the AWS cloud and on Amazon Rekognition, which allowed the assessment to be completely online.

The key points of the provided solution are:

  • Identity validation via webcam
  • Cloud architecture
  • Integration into existing assessment processes
  • Reports for assessors

Identity validation tools for online assessments are based on the comparison of images captured during the test with a reference image. In the system used in this case, the validation is performed against an initial set of images, with different angles of the face of the person taking the exam.

This set of images is prepared at the beginning of the semester, and it is validated against the image of the identity document available at the university.

During the exams, the system randomly captures images of the person being assessed. These images are analyzed, looking for unusual situations (different people taking the exam, more than one person in front of the computer, absences of the assessed person, etc.).

Validación identidad evaluación online Amaron Rekognition

The architecture of this system is based on the AWS cloud platform, making use of Amazon Rekognition, its computer vision platform.

This architecture was designed considering the enrollment and assessment processes according to the university's requirements. For its dimensioning, we analyzed the volumes and peaks derived from its exam schedule, being able to analyze more than 4000 tests simultaneously.

Amazon Rekognition_arquitectura cloud_validación identidad_evaluación online

At backoffice and Business Intelligence levels, different Grafana panels provide real-time information during the realization of the exams. The system can identify those students having technical problems, which allows the university's IT services to help solve them.

Amazon Rekognition_backoffice_validación identidad_evaluación online

The output of the assessment process is a report with the results of the images analyzed in each exam.

A traffic light system is also provided, so that only those exams with issues need to be reviewed in order to validate any attempt of cheating.

Results

In 6 days of exams in January 2022:

  • 70,000 exams were taken with the identity validation system for online assessments.
  • 9,000,000 images were compared by AWS Rekognition.
  • 26% of electronic supervision reports were manually reviewed (target <50%).

This prestigious online university choose Opentrends as its partner for this project due to:

  • Our experience in large projects for the education sector.
  • Our skills for developing architectures in the AWS cloud.
  • Our ability to tackle projects with a high degree of innovation.
Contact us

If you are looking for a partner that helps you develop the identity validation solution you need, just contact us.