MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641058069 A) filed by The Principal, Mepco Schlenk Engineering College on May 07, 2026, for Coercion Aware Biometric Vault With Integrated Emotion Intelligence.
Inventors include Dr. S. Selva Nidhyananthan; C. Karuppaiah; and K. Kavi Bala.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: Coercion-Aware Biometric Vault with Integrated Emotion Intelligence Coercion-Aware Biometric Vault with Integrated Emotion Intelligence Abstract: The present invention discloses a Coercion-Aware Biometric Vault with Integrated Emotion Intelligence, providing a multimodal security system designed to detect and prevent unauthorized access during duress or coercion scenarios. As illustrated in the enclosure setup of Fig. 2 and the prototype in Fig. 5, the system is housed in a vault body (1) comprising a front control panel (2) and a main locker door (8) secured by a solenoid lock (6). The entire logic is managed locally by a processing unit (13), specifically a Raspberry Pi 5, powered by a 5V power supply (14). As depicted in the logic flow of Fig. 1 and the architectural modules of Fig. 4, the authentication process initiates when a camera (3), connected via a camera cable (15), captures a live video stream of the user (10). The processing unit (13) first executes a Face Verification (21) module. This includes a aliveness check using the Eye Aspect Ratio (EAR) algorithm to detect physical eye blinking, followed by Face Recognition (17) using a ResNet-34 architecture to match the user against a stored database. Upon successful identity verification, the LCD display (4) provides personalized feedback. Simultaneously, the system performs Emotion Analysis (18) using the HSEmotion model (EfficientNet-BO) to scan for the biological signature of "Fear." As seen in Fig. 1, if the averaged fear score over a sliding window exceeds a preset safety threshold, the system triggers a "Safety Override" protocol. In this state, the processing unit (13) activates a buzzer (7) to sound an alarm and maintains the solenoid lock (6) in a locked state, effectively denying access even to an authorized user who is under threat. If the user is determined to be in a safe emotional state (neutral or happy), the system prompts for a Voice Signal (20) via the microphone (5). A Speaker Recognition ( 19) module, utilizing the ECAP A-TDNN architecture, verifies the vocal tract resonance using cosine similarity. Only upon the successful tripartite validation of identity, safe emotion, and voice, the processing unit (13) triggers the 5V relay module (11 ). This bridges the 12V power supply (12) through a connector (16) to retract the solenoid lock (6), granting access. The LCD display (4) subsequently updates with the specific locker number (9) and access status. All biometric computations are performed at the edge on the processing unit (13), ensuring maximum data privacy and low latency response.
Disclaimer: Curated by HT Syndication.
