Loading...
Please wait, while we are loading the content...
Similar Documents
Efficiently Learning a Detection Cascade with Sparse Eigenvectors
| Content Provider | arXiv |
|---|---|
| Author | Shen, Chunhua Paisitkriangkrai, Sakrapee Zhang, Jian |
| Date of Submission | 2009-03-18 |
| Abstract | In this work, we first show that feature selection methods other than boosting can also be used for training an efficient object detector. In particular, we introduce Greedy Sparse Linear Discriminant Analysis (GSLDA) \cite{Moghaddam2007Fast} for its conceptual simplicity and computational efficiency; and slightly better detection performance is achieved compared with \cite{Viola2004Robust}. Moreover, we propose a new technique, termed Boosted Greedy Sparse Linear Discriminant Analysis (BGSLDA), to efficiently train a detection cascade. BGSLDA exploits the sample re-weighting property of boosting and the class-separability criterion of GSLDA. |
| Related Links | https://arxiv.org/pdf/0903.3103.pdf |
| Page Count | 12 |
| arXiv | 0903.3103 |
| Language | English |
| Access Restriction | Open |
| Subject Keyword | Computer Science - Multimedia Computer Science - Artificial Intelligence Computer Science - Machine Learning Computer Science |
| Content Type | Text |
| Resource Type | Article |
| Subject | Computer Science |
