• Corpus ID: 56227347

A Literature Study on Crowd (People) Counting With the Help of Surveillance Videos

@inproceedings{Kowcika2015ALS,
  title={A Literature Study on Crowd (People) Counting With the Help of Surveillance Videos},
  author={A. Kowcika and S.Sridhar S.Sridhar},
  year={2015}
}

• A. Kowcika, S. S.Sridhar
• Published 12 August 2015
• Computer Science

The categories of crowd counting in video falls in two broad categories: (a) ROI counting which estimates the total number of people in some regions at certain time instance (b) LOI counting which counts people who crosses a detecting line in certain time duration. The LOI counting can be developed using feature tracking techniques where the features are either tracked into trajectories and these trajectories are clustered into object tracks or based on extracting and counting crowd blobs from a temporal slice of the video. And the ROI counting can be developed using two techniques: Detection Based and Feature Based and Pixel Regression Techniques. Detection based methods detect people individually and count them. It utilizes any of the following methods:- Background Differencing, Motion and Appearance joint segmentation, Silhouette or shape matching and Standard object recognition method. Regression approaches extract the features such as foreground pixels and interest points, and vectors are formed with those features and it uses machine learning algorithms to subside the number of pedestrians or people. Some of the common features according to recent survey are edges, wavelet coefficients, and combination of large set of features. Some of the common Regressions are Linear Regression, Neural Networks, Gaussian Process Regression and Discrete Classifiers. This paper aims at presenting a decade survey on people (crowd) counting in surveillance videos.