Smart Cameras and Neural Networks to Help Identify Cheating Students in Sverdlovsk Region During Unified State Exam

The system automatically detects violations and signals the observer

In the Sverdlovsk region, a neural network has been implemented for the early Unified State Exam (USE) to analyze participants' behavior. As reported by Rostelecom, upon detecting prohibited items or attempts to remove materials, the system automatically signals, records the violation, and sends the recording for observer review. After moderator approval, the data proceeds for further processing.

Each examination point is equipped with at least two high-resolution IP cameras with quality sound – a total of 16 devices for 124 early period participants. The system operates on secure communication channels and includes IP cameras, access switches, vandal-proof cabinets, and uninterruptible power supplies. Digital cameras cover the necessary viewing angle: participants, printing, scanning, and material distribution areas are captured, while personal data, form content, and answers remain confidential.

Video from the classrooms is broadcast to the "Smotri EGE" (Watch USE) portal in real-time, with simultaneous access for 15,000 accredited specialists and public observers.

At the federal level, Rostelecom has equipped 1,600 classrooms in 68 regions of Russia – a total of 3,100 devices, ensuring 100% coverage of classrooms. The archive of video recordings will be stored in Rostelecom's data centers until March 1, 2027; for additional reliability, each camera is equipped with a local memory card.

Read more on the topic:

Now on home