MUMBAI, India, March 13 -- Intellectual Property India has published a patent application (202641025580 A) filed by Dayananda Sagar College Of Engineering, Bangalore, Karnataka, on March 4, for 'a knowledge-driven ontology-integrated detr architecture for human-object interaction recognition.'
Inventor(s) include Rashmi S R; Dr. Anusha Preetham; and C Manasa.
The application for the patent was published on March 13, under issue no. 11/2026.
According to the abstract released by the Intellectual Property India: "This invention reveals a Detection Transformer supported model with a Knowledge driven Ontology Layer to Human Object Interaction (HOI) recognition. The system combines both convolutional transformer hybrid model with knowledge driven ontology model to enhance accuracy and flexibility of HOI detection. First, feature extraction and transformer layers are used to process first video frames to recognize humans, objects and their interactions. The identified entities and their interactions are then mapped into a domain ontology in which a semantic knowledge and rule-based inferences are implemented as description logic and SWRL rules. This ontology-based reasoning allows higher level interpretations that were overlooked in the former step. The algorithm eliminates the reliance on huge training data that is annotated. This also makes sure that there is explainable decision making and generalization between domains. Video surveillance, Robotics, autonomous systems, AR/VR, human-computer interaction and healthcare monitoring are general applications."
Disclaimer: Curated by HT Syndication.
