Display Reflection Detection for Deepfake-Resistant Presence Checks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing online personal presence verification systems are vulnerable to deepfakes and cheating methods, such as pre-recorded videos or unauthorized assistance, which compromise the integrity of online examinations and digital security.
Innovation Solution
A system and method utilizing reflection detection and machine learning models to measure changes in brightness and color temperature on a user's face in response to displayed objects, verifying the presence of the individual in front of the camera by comparing these changes to predefined thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional camera-based presence verification is used, then the system is simple to operate, but it is vulnerable to deepfakes and pre-recorded videos
Solution Approach 1:
The system dynamically changes display parameters (brightness, color temperature) in real-time and monitors corresponding changes in facial reflections. This dynamic approach transforms static presence verification into an active detection process, making it resistant to pre-recorded videos while maintaining operational simplicity through automated parameter adjustment and analysis
Solution Approach 2:
The invention changes physical parameters of the display (brightness levels, color temperature) and measures the corresponding changes in reflected light on the user's face. By monitoring these parameter changes rather than static images, the system achieves higher reliability in detecting genuine presence versus deepfakes or recorded videos
2Measurement precision
If reflection detection with multiple parameters is implemented, then detection precision is improved, but measurement complexity increases
Solution Approach 1:
The system replaces complex manual analysis of facial reflections with automated machine learning models that process brightness and color temperature data. This substitution of mechanical/manual measurement with computational analysis maintains high detection precision while reducing the practical difficulty of implementation
Solution Approach 2:
The invention introduces machine learning models as intermediaries between the raw facial reflection data and the final presence verification decision. These models automatically extract meaningful patterns from brightness and color temperature changes, simplifying the measurement process while maintaining high precision in detecting genuine user presence
3Productivity
If automated proctoring with AI is used, then productivity is improved, but reliability decreases due to cheating vulnerabilities
Solution Approach 1:
The system performs preliminary presence verification by analyzing facial reflections before and during the examination process. By establishing baseline reflections and continuously monitoring for deviations, the system proactively detects cheating attempts rather than reacting after they occur, maintaining both high productivity and reliability
Solution Approach 2:
The invention implements continuous feedback loops where the system monitors facial reflection changes in real-time and compares them against expected patterns. When deviations are detected (indicating potential cheating), the system can trigger alerts or terminate the examination, ensuring reliability while maintaining automated high-speed processing
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Effectively distinguishes between the genuine user and impostors by detecting discrepancies in brightness and color temperature, thereby preventing cheating and ensuring the authenticity of online examinations.
Implementation Method 1
obtaining changes in a brightness and a color temperature of at least one surface on a face of the user from the video stream; based on a determination that the obtained changes in the brightness and color temperature of surfaces on a face of the user correspond to the brightness and color temperature of the object
Data Source
AI summary
Disclosed herein are systems and methods for detecting a presence of a person in front of a display with a camera based on a reflection detection. In one aspect, an exemplary method includes obtaining, using a camera, a video stream of a user in front of the display. The method also includes changing a brightness and color characteristics of an object. The method further includes obtaining changes in a brightness and a color temperature of at least one surface on a face of the user from the video stream. The method further includes based on a determination that the obtained changes in brightness and color temperature of surfaces on the face of the user do not correspond to the brightness and color temperature of the object within the brightness threshold and the color temperature threshold, determining that the user is not in front of the display.