MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202641093678 A) filed by Prathyusha Engineering College on August 03, 2026, for Real-Time Hand Landmark Extraction And Cnn- Based Stylistic Classification For Artistic Overlays.
Inventors include Ms. V. Asha; Ms. Sabasri K; Ms. Pavithra Sb; Ms. Priyanka B; Ms. Jothi Lakshmi; Ms. P. Sivaganga; Ms. R. Kannamma; Dr. C. Srivenkateswaran; Mr. Samuel Dinesh Hynes; and Ms. B. Angalaparameswari.
The application for the patent was published on August 07, 2026, under issue no. 32/2026.
Abstract: The present invention provides a comprehensive system for high-fidelity, real-time stylistic transformation of hand-based video inputs. The process commences with the continuous acquisition of raw image data from an imaging device, referred to as the Video Input (Data Acquisition), which provides a stable, low-latency stream suitable for subsequent processing. Upon capturing these frames, the system initiates the Hand Landmark Extraction (Feature Detection) phase, wherein advanced machine learning models, specifically utilizing coordinate-based pose estimation, detect and isolate specific anatomical landmarks in three-dimensional space, thereby mapping the spatial orientation and dynamic posture of the user's hand. Following successful landmark identification, the cropped hand imagery is passed through a Stylistic Classification (CNN Model), which employs a deep neural network—trained on a diverse dataset of artistic representations—to analyze texture and contour, ultimately classifying the image into a specific stylistic category such as watercolor, cartoon, sketch, or oil painting. The output from the classification stage, in conjunction with the dynamic landmark data, is then utilized in the Artistic Overlay Generation & Rendering module, which synthesizes the requested artistic effects and applies them in precise alignment with the detected hand landmarks to ensure fluid, jitter-free visual coherence. Finally, the processed signal is pushed to the Final Real-Time Output interface, delivering a seamless augmented reality experience wherein the user's hand is displayed in the chosen artistic medium in real-time, maintaining high frame rates required for responsive interaction.
Disclaimer: Curated by HT Syndication.
