|Title:||Pedestrian detection in crowded scenes|
|Subject:||Pedestrian traffic flow.|
Pedestrians -- Mathematical models.
Image analysis -- Mathematical models.
Hong Kong Polytechnic University -- Dissertations
|Department:||Department of Electronic and Information Engineering|
|Pages:||v, 74 p. : ill. ; 30 cm.|
|Abstract:||Pedestrian detection has attracted a lot of attention recently, as it is important for applications such as home security, video surveillance, vehicle-assistance systems, crowd control, etc. In this thesis, we have studied different efficient features for representing pedestrians; these include histograms of oriented gradients (HoG), Haar-like features, region covariance, etc. Based on these features, classifiers or detectors can be designed to detect pedestrian objects. We have investigated the support vector machine (SVM) and the AdaBoost algorithm. SVM can achieve a high detection rate, but this method is more computational than the AdaBoost method. Therefore, we first implement and evaluate the performance of the SVM method and the AdaBoost method separately. Then, we propose to combine these two approaches so that the overall algorithm can achieve optimal efficiency as well as accuracy. In our method, we first apply the AdaBoost algorithm so that most of the non-pedestrian candidates can be removed efficiently. Then, the remaining possible candidates are verified by using the SVM method. We compare the three different approaches for pedestrian detection, and we found that our proposed combined method can achieve the best performance.|
|Rights:||All rights reserved|
Files in This Item:
|b24757652.pdf||For All Users (off-campus access for PolyU Staff & Students only)||1.32 MB||Adobe PDF||View/Open|
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item: