MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202631089457 A) filed by National Institute Of Technology Silchar on July 22, 2026, for Synchronized Multi-Angle Facial Analysis System And Method For Subject Recognition, Pose- Estimation, And Facial Landmarks Detection.
Inventors include Dr. Kuldeep Singh Yadav; Ms. Masuma Aktar; Prof. Rabul Hussain Laskar; and Prof. Prof. Fazal A. Talukdar.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: The invention relates to a synchronized multi-angle facial analysis system and method are disclosed for simultaneous subject recognition, pose estimation, and facial landmark detection. The system comprises a semi-circular frame(102) supporting multiple synchronized cameras(104), an adjustable chair(106) to centre a subject's head, and variable illumination sources(108) configured to project diverse lighting conditions via distinct topological arrangements. Operationally, a data input unit(110) receives captured multi-angle images to generate a resolution-invariant composite input tensor. A feature extraction unit(112), utilizing a sequence of inverted residual blocks combined with a hybrid attention mechanism, processes this tensor to yield a refined feature map. Finally, a controlling unit(114) utilizes multiple distinct linear projections to map the refined feature map into concurrent face-related predictions. Specifically, the system executes simultaneous multi-tasking to predict subject identity using a non-linear activation function, multi-axis face orientation across predefined angle dimensions, and coordinate pairs for localized facial key points.
Disclaimer: Curated by HT Syndication.
