Web Page Investigator for Proactive QR Code Malware Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional QR code scanner apps require users to scan QR codes to identify potential malicious links, and there is no proactive solution to warn users about malicious QR codes before scanning.
Innovation Solution
A computer vision model is used to scan snapshots of web pages to identify and annotate graphical features, including QR codes, before rendering the content. The model extracts information from identified QR codes and transmits it to a phishing and content protection (PCP) engine for assessment, providing an indication of potential maliciousness.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional QR scanner apps are used to identify malicious links, then users can be warned about dangerous QR codes, but users must first scan the QR code which exposes them to potential malicious websites
Solution Approach 1:
The system performs preliminary scanning of QR codes on web pages using computer vision models before users can scan them. The security assessment happens in advance, identifying malicious QR codes and preventing user exposure before the harmful interaction occurs.
Solution Approach 2:
The patent introduces an intermediary security system that sits between the user and the QR code scanning process. The computer vision model and PCP engine act as intermediaries to assess QR codes automatically, preventing direct user exposure to potentially malicious content.
2Reliability
If QR codes are scanned to identify malicious content, then security assessment can be performed, but the scanning process itself may lead to malware transmission
Solution Approach 1:
The system performs malware detection through computer vision analysis of QR code visual patterns before any scanning or transmission occurs. The assessment is based on visual feature extraction and pattern recognition, eliminating the need for actual QR code scanning that could trigger malware execution.
Solution Approach 2:
The patent replaces the traditional mechanical scanning process with computer vision-based visual analysis. Instead of scanning QR codes which may trigger malware, the system uses image processing and neural networks to analyze visual patterns, substituting a safe analytical method for a potentially harmful one.
3Productivity
If digital QR codes are displayed on web pages, then information can be transmitted efficiently, but users cannot know if the QR code is malicious before scanning
Solution Approach 1:
The system performs preliminary security assessment of QR codes on web pages before users interact with them. The computer vision model scans and evaluates QR codes in advance, and security status information is prepared and ready for display, preventing information loss about malicious content.
Solution Approach 2:
The patent implements a feedback mechanism where the PCP engine provides security status information back to the system, which then displays warnings or indicators to users. This feedback loop ensures that security status information is not lost but actively communicated to users before they scan QR codes.
Data Source
AI summary
Methods and systems for annotating and assessing content, including QR codes, using a web page investigator to avoid dangerous scans of such content are described herein. A security module executing on a client device may receive a request for web content. A computer vision model may then scan the requested web content to identify and annotate graphical features on the webpage prior to rendering the web content on a display of the client device. The computer vision model may identify a QR code and transmit information encoded within the QR code to a server executing a phishing and content protection (PCP) engine for analysis. When the identified QR code is indicated to be malicious, the client device may render a modified version of the requested web content to discourage the user from scanning the identified QR code.