Learning to localize detected objects
Authors:
- 1 June 2012
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 10636919,p. 3322-3329
Abstract
In this paper, we propose an approach to accurately localize detected objects. The goal is to predict which features pertain to the object and define the object extent with segmentation or bounding box. Our initial detector is a slight modification of the DPM detector by Felzenszwalb et al., which often reduces confusion with background and other objects but does not cover the full object. We then describe and evaluate several color models and edge cues for local predictions, and we propose two approaches for localization: learned graph cut segmentation and structural bounding box prediction. Our experiments on the PASCAL VOC 2010 dataset show that our approach leads to accurate pixel assignment and large improvement in bounding box overlap, sometimes leading to large overall improvement in detection accuracy.Keywords
This publication has 17 references indexed in Scilit:
- Semantic contours from inverse detectors2011
- The truth about cats and dogs2011
- Image segmentation with a bounding box prior2009
- Multiscale Categorical Object Recognition Using Contour FragmentsIEEE Transactions on Pattern Analysis and Machine Intelligence, 2008
- Using contours to detect and localize junctions in natural images2008
- Using Segmentation to Verify Object Hypotheses2007
- A Statistical Approach to Texture Classification from Single ImagesInternational Journal of Computer Vision, 2005
- Pedestrian Detection in Crowded Scenes2005
- "GrabCut"2004
- Support vector machine learning for interdependent and structured output spaces2004