A method for recognizing dense crowds in scenic spots based on a LUV color model

By converting RGB images to LUV images and constructing a Mamba network, combined with data augmentation and multi-module feature extraction, the robustness and real-time performance issues of dense crowd identification in scenic areas under complex lighting and backgrounds were solved, achieving more accurate dense crowd identification and monitoring.

CN122157138APending Publication Date: 2026-06-05XIAN TECH UNIV +1
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
XIAN TECH UNIV
Filing Date
2026-01-07
Publication Date
2026-06-05

AI Technical Summary

Technical Problem

Existing methods for identifying dense crowds in scenic areas lack robustness and real-time performance under complex backgrounds and varying lighting conditions, making it difficult to effectively identify and monitor dense crowds.

Method used

The Mamba network based on the LUV color model is used to convert RGB images to LUV images, combine random cropping, flipping and rotation for data augmentation, and construct the Mamba network, including shallow convolutional feature extraction, deep semantic feature extraction, multi-branch fusion, multi-scale convolution and spatial attention module, to generate prediction boxes to identify dense crowds.

Benefits of technology

It improves the recognition accuracy and robustness in complex lighting environments, reduces the impact of background interference, and achieves more accurate recognition and real-time monitoring of dense crowds.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122157138A_ABST
    Figure CN122157138A_ABST
Patent Text Reader

Abstract

The present application relates to a kind of Mamba network scenic spot crowded people identification method, specifically to a kind of Mamba network scenic spot crowded people identification method based on LUV color model.The method includes the following steps:1, the original RGB image to be identified is converted into LUV image;2, LUV image is sequentially subjected to random cutting, random flip and random rotation, realizes data enhancement;3, the LUV image after data enhancement is divided into training set and test set, and constructs Mamba network;4, the LUV image in training set is input into Mamba network training, obtains the Mamba network of training;5, the LUV image in test set is input into the Mamba network of training and generates prediction frame, i.e. the density detection frame for indicating crowded area position.The present application can effectively reduce the influence of illumination variation on crowded people detection, improve robustness and accuracy in crowded people identification task, especially suitable for complex illumination environment under scenic spot scene.
Need to check novelty before this filing date? Find Prior Art