Project overview

The system relies on a customized version of the OpenCV library to ensure dependable student identity verification as well as effective exam proctoring — with features like face recognition, light level, distance from camera, head movement and headgear detection, and more.

Client: an established U.S. leader in exam monitoring dedicated to helping educational institutions around the world protect their online programs’ academic integrity. The client’s signature solution is designed to meet the highest standards of the traditional proctoring experience while enabling students and professionals to take tests from a location most convenient to them.

Delivery Model
SLA-driven task backlog delivery (TM-based)
Effort and Duration
9 months, 36 man-months
React • MobX • axios • C++ • OpenCV • Ruby on Rails • PostgreSQL • ImageMagick • FreeTDS • Java • Oracle • Kurento Media Server • Amazon S3

Business challenge

The customer, a top-rated U.S.-based proctoring firm serving students from 129 countries, needed to improve their business-critical test and exam monitoring solution in several key areas.

Based on Adobe Flash, the system was relying on a soon-to-be-phased-out technology and needed to transition to a modern media communication stack in order to retain a competitive edge.

Another major pain point was the need to supercharge monitoring efficiency by introducing automation into the proctoring process. By relying on smart, data-driven automation, the client would eventually expand its capacity by an estimated 93% — all without hiring additional staff.

Oxagile's differentiators

Oxagile established its credibility as the best vendor to handle the project owing to the following competitive advantages:

Proven WebRTC know-how.

Migrating to a modern real-time communication tech stack called for a vendor with a strong WebRTC background, including in-the-trenches experience in working with Kurento Media Server.

Domain-specific knowledge.

Because the client’s platform tightly integrates with LMS products such as Blackboard and Canvas, good understanding of the EdTech space on the part of Oxagile helped speed up both requirements management and the implementation of the LTI-related functionality.

Solid C++ and OpenCV expertise.

Automating session launching and exam proctoring required delving into the bleeding-edge of technology, including machine learning-assisted face recognition. This is where Oxagile’s C++ and OpenCV expertise proved invaluable.

Predictable delivery process.

Some of the project’s loosely defined requirements, as well its sheer complexity and innovative nature carried a number of technological and process risks. Making the project work required a time-tested, predictable Scrum-based workflow that would ensure on-time and on-budget delivery.

For this project, Oxagile put together a versatile team that included a Scrum Master, several senior Ruby and JavaScript professionals, a veteran C++ programmer, a Java developer, a UI designer, as well as the necessary testing and QA specialists who got involved on an as-needed basis.

Delivered solution

The resulting web solution is a top-grade proctoring platform that combines advanced automation technology with the human element to provide a dependable way of overseeing hundreds of simultaneous remote exams anywhere in the world.

The solution leverages a modern, future-proof WebRTC-based tech stack for video recording, features machine learning-powered authentication, and relies on data-driven automation to intelligently flag suspicious activity on the part of the test-taker.

Key features and highlights

  • Intuitive exam monitoring, with support for both live and automated sessions
  • Seamless WebRTC-based video stream capture and playback (both the test-taker’s face and their desktop actions are recorded)
  • Automated session flagging with a wide range of red flag triggers (light and audio level changes, head movements, the appearance of unauthorized persons in the room, window resize events, and more)
  • Scoring functionality that assigns penalty points for suspicious activity and notifies watchers
  • Challenge question authentication that works by pulling publicly available data and asking a student to do a short quiz
  • Sophisticated user identity validation based on a biometric profile (keystroke/typing style analysis)
  • Innovative machine learning-supported face recognition based on a customized build of the OpenCV library
  • Easy-to-use text and audio chat functionality to facilitate communication between proctors and students
  • Advanced reporting and statistics across test-takers, exams, and educational establishments

From a technology perspective, the system encompasses a range of modules, including a Google Chrome extension for exam watchers and test-takers, a Java-based backend, a Kurento-powered media streaming server (enhanced with a custom C++/OpenCV-based plugin), and an advanced exam scheduling subsystem implemented in Ruby.

Project results

By partnering with Oxagile, the client received a much-improved proctoring solution designed to effortlessly monitor 5,000+ daily exams via both live and automated sessions. In the near future, the number of sessions is expected to double.

The new platform’s efficiency gains will allow the client to scale their business at an order of magnitude faster rate, outstripping the competition and reinforcing their position of a well-established proctoring service provider.

Currently work is underway to further enhance the platform in areas such as reporting, score calculation and comparison, smart audio and ambient light baseline identification, IP and geolocation change detection, and much more.

After the initial release, the client plans to leverage Oxagile’s expertise in performance testing and optimization to monitor for and eliminate any performance bottlenecks and implement professional-quality autoscaling.


To get your project underway, simply contact us and an expert will get in touch with you as soon as possible.