Combining Weighted Templates with HOGs for Pedestrian Detector Training
| Autor: | Shih-han Ku, 顧詩涵 |
|---|---|
| Rok vydání: | 2013 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 101 Detecting pedestrian is an important step in many areas, such as intelligent transportation systems (ITSs) or visual surveillance. Currently, boosting a set of local features based on histogram of oriented gradients (HOGs) to form a pedestrian detector has proven its effectiveness in the literature. However, this kind of approaches suffer from the problem of false detection in case of complex background or noise effect. Accordingly, the main objective of this work is to alleviate this problem by incorporating the results from the template matching into the boosting framework. To achieve this, the idea behind is to tune the hyper-plane of the support vector machine according to the template-based classifier at each round of boosting stage so that the formed detector has a bias against false positive. Instead of manual annotation, a set of representative templates are automatically generated based on expectation maximization (EM) by giving a set of training samples. To make the template have more discriminative power, each point in the generated binary-contour template is assigned to a different importance (weight) but not considered as equal in matching. The experiments provided exhibit the superiority of the proposed method in detection accuracy. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|