MUMBAI, India, May 1 -- Intellectual Property India has published a patent application (202611027396 A) filed by Manipal University, Jaipur, Rajasthan, on March 9, for 'a system and method for adaptive artificial intelligence-based crowd headcount estimation using density-based model switching.'

Inventor(s) include Dr. Aprna Tripathi; Jashvant Kumar; Vinay Agrawal; and Dr. Usha Jain.

The application for the patent was published on May 1, under issue no. 18/2026.

According to the abstract released by the Intellectual Property India: "The present invention relates to a system and method for adaptive crowd headcount estimation using artificial intelligence with density-based model switching. The system comprises a data acquisition module to take crowd images through digital cameras; a preliminary density assessment module performs an initial estimation of crowd density within the captured scene; an adaptive model switching module dynamically selects and activates an appropriate artificial intelligence model depending on the detected crowd density scenario; an output module to display the final headcount. The system further comprises a cloud server to store the crowd images. Reliable people counting are still a challenging issue in computer vision, since crowd density, scale, occlusion and perspective changes severely affect the counting rates. Nearly all conventional solutions using only object detection or density estimation do not generalize well to various scenes. To cope with this deficiency, we design a dynamic hybrid framework that can dynamically switch between two models which complement each other: a discrete object detection network (YOLOv8) for sparse crowd scenarios and a density map estimation network (CSRNet) for dense crowd scenarios. The system also incorporates a density-aware switching strategy, which can automatically choose a proper model according to crowd situation, thereby increasing the accuracy and reducing the computational cost at the same time. Real-world applications such crowd monitoring in the field of public safety, event management, urban planning, etc. can be done with such a unified pipeline, offering a scalable, efficient and accurate solution to crowd counting that may trade off the accuracy of the detection with the computational cost."

Disclaimer: Curated by HT Syndication.