Optimizing the cost function of histogram of oriented gradient-based INRIA dataset
Person detection in images requires both image processing and machine learning concepts. Image processing techniques are used in extracting feature descriptor sets. The extracted features are then used as inputs for training a machine learning algorithm to perform classification of objects are perso...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | text |
| Published: |
Animo Repository
2014
|
| Subjects: | |
| Online Access: | https://animorepository.dlsu.edu.ph/faculty_research/12626 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | De La Salle University |
| Summary: | Person detection in images requires both image processing and machine learning concepts. Image processing techniques are used in extracting feature descriptor sets. The extracted features are then used as inputs for training a machine learning algorithm to perform classification of objects are persons. One of the feature description algorithms used for image classification is the Histogram of Oriented Gradients (HOG). HOG is based on gradient vectors and the use of sliding windows in order to obtain the feature descriptor sets. For machine learning, support vector machine (SVM) is used for person classification. In this paper, the images used are based on the INRIA person dataset, which contains 3542 human images with varying range of pose and backgrounds. This paper presents the finding of the optimized cost function C for each type of linear-based SVM models, for person detection in the INRIA person data set, based on the HOG feature detector set. |
|---|