Please use this identifier to cite or link to this item:
|Title:||Extreme learning machine for salient object detection in images|
|Citation:||WMSCI 2018 - 22nd World Multi-Conference on Systemics, Cybernetics and Informatics, Proceedings, 2018, Vol.1, , pp.39-44|
|Abstract:||Salient areas within an image are natural points of focus for a typical human eye. Identification of these areas is a key step for further applications like object detection, segmentation, recognition and so on. We aim to develop a fast and accurate learning based model for image saliency detection. We use learning based method to replace the traditional contrast based low level feature oriented methods. In this paper, an effective technique which makes use of Extreme Learning Machine (ELM) for detection of salient areas within an image is proposed. These salient areas are identified at the superpixel level. We use the existing methods which moderately detect salient areas to generate prior maps. Through these, training samples are collected in order to train the classifier. We present a detailed comparison of features used for training the classifier and an optimal set of features that should be used for training for improving the state of the art methods. Copyright 2018. � by the International Institute of Informatics and Systemics.|
|Appears in Collections:||2. Conference Papers|
Files in This Item:
There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.