MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202641091733 A) filed by Akshaya College Of Engineering And Technology on July 29, 2026, for Violence Detection In Crowd Using Openai.
Inventors include Ms. K. Lavanya; Yeddulaganesh; Palukurunagakarthik; Tadipatrisaicharankumar Reddy; and Rubanraj S.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: ABSTRACT Violence Detection In Crowd Using OpenAI With the rapid growth of urban populations and large public gatherings, ensuring safety in crowded environments has become a major challenge. This paper presents an intelligent and scalable violence detection framework that identifies suspicious and violent activities in real time using image and video data. The system is built on the CLIP (Contrastive Language– Image Pretraining) vision–language model, enabling semantic understanding by mapping visual content and textual descriptions into a shared feature space, which supports zero-shot classification without requiring extensive task-specific datasets and allows detection of events such as physical fights, fires, road accidents, and other abnormal behaviors. An interactive user interface developed using Streamlit enables users to upload images or videos for instant analysis, while an optimized frame sampling strategy reduces computational load during video processing without compromising accuracy, making the system suitable for resource-constrained environments without GPUs. Each frame is analyzed individually and aggregated to determine overall context, and an automated alert mechanism sends email notifications via Gmail with visual evidence when potential threats are detected, ensuring rapid response. The modular design allows seamless integration with existing surveillance systems like CCTV networks, and experimental results show reliable performance across varying conditions, demonstrating improved generalization over traditional supervised methods, making the system a cost- effective, scalable, and efficient solution for public safety monitoring and emergency response.
Disclaimer: Curated by HT Syndication.
