Passenger Flow Monitoring With Two-Level Video Scene Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing intelligent analysis systems in subway stations struggle with high resource consumption, inadequate passenger flow monitoring, and manual guidance due to complex environments, leading to inefficient crowd dispersion and safety risks.
Innovation Solution
A method and system for estimating and presenting passenger flow using two-level scene classification, automatic algorithm configuration, and a base layer characteristic sharing model to analyze on-site video data, reducing resource usage and enabling real-time monitoring and guidance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the intelligent analysis system accesses more cameras for comprehensive passenger flow monitoring, then the measurement precision of passenger flow is improved, but the server resource occupation increases
Solution Approach 1:
The patent divides the video processing task into two stages: first, a lightweight algorithm performs initial analysis on all camera feeds to identify scenes with passenger flow; second, the full intelligent analysis system processes only those specific video segments where passengers are present. This segmentation reduces overall server resource consumption while maintaining comprehensive monitoring coverage across all cameras.
Solution Approach 2:
The system applies full intelligent analysis algorithms selectively rather than continuously on all camera feeds. By using partial action (processing only relevant video segments) rather than excessive action (continuous full analysis), the system achieves accurate passenger flow detection where needed while minimizing unnecessary server resource occupation during periods without passengers.
2Adaptability or versatility
If manual algorithm adaptation and ROI delimitation are performed for each subway station, then the adaptability of the system to different stations is improved, but the deployment complexity and manpower requirements increase
Solution Approach 1:
The system automatically adapts to different subway station environments through self-service mechanisms. The lightweight algorithm automatically identifies scenes containing passengers and delimits ROI areas without requiring manual configuration. This automated adaptation process eliminates the need for professional technicians to manually adjust algorithms for each station, significantly reducing deployment complexity while maintaining high adaptability to different station characteristics.
Solution Approach 2:
The system performs preliminary automatic scene classification and ROI identification during the deployment phase, preparing configuration data in advance. This preliminary action allows the system to be pre-adapted to specific station layouts and characteristics before actual operation, reducing the need for manual adjustments later while ensuring station-specific adaptability.
3Device complexity
If simple broadcast guidance is used for passenger flow management, then the system complexity is reduced, but the effectiveness of crowd dispersion and passenger information provision deteriorates
Solution Approach 1:
The system implements a feedback mechanism where real-time passenger flow data from video analysis is continuously monitored and used to dynamically adjust guidance information. When passenger flow exceeds thresholds or congestion is detected, the system automatically generates and displays targeted guidance messages on station screens, providing timely feedback to passengers about crowd conditions and suggesting alternative routes or waiting areas, thereby improving crowd dispersion effectiveness without requiring complex manual intervention.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A method for estimating and presenting a passenger flow, a system, and a computer storage medium. The method comprises: performing two-level scene classification on received on-site video data transmitted by a camera; configuring a passenger flow analysis algorithm corresponding to a scene according to a scene classification result; and analyzing a video frame according to the passenger flow analysis algorithm, outputting a passenger flow calculation result, determining, according to the passenger flow calculation result, a crowdedness level of a passenger flow at a location corresponding to the camera, and respectively transmitting the crowdedness level of the passenger flow to a first terminal and a second terminal for display of the crowdedness level of the passenger flow.