Abstract
An industrial neural network based crowd monitoring system for surveillance at underground station platforms is presented. The developed system was thoroughly off-line tested by video images obtained from the underground station platform at Hong Kong. The developed system enables the density level of crowd to be automatically estimated. Crowd estimation is carried out by extracting a set of significant features from sequence of video images. The extracted features are modelled by a neural network for estimating the level of crowd density. The learning process is based upon an efficient hybrid type global learning algorithms, which are capable of providing good learning performance. Very promising results were obtained in terms of estimation accuracy and real-time response capability to alert the operators automatically.
Original language | English |
---|---|
Pages (from-to) | 73-83 |
Number of pages | 11 |
Journal | Advanced Engineering Informatics |
Volume | 16 |
Issue number | 1 |
DOIs | |
Publication status | Published - Jan 2002 |
Externally published | Yes |
Keywords
- Crowd estimation
- Hybrid global learning algorithm
- Neural networks
- Underground station platform
ASJC Scopus subject areas
- Information Systems
- Artificial Intelligence