Handling Occlusions with Franken-Classifiers
Authors:
- Markus MathiasMMMarkus Mathias
- Rodrigo BenensonRBRodrigo Benenson
- Radu TimofteRadu Timofte
- Luc Van GoolLGLuc Van Gool
- 1 December 2013
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 15505499,p. 1505-1512
Abstract
Detecting partially occluded pedestrians is challenging. A common practice to maximize detection quality is to train a set of occlusion-specific classifiers, each for a certain amount and type of occlusion. Since training classifiers is expensive, only a handful are typically trained. We show that by using many occlusion-specific classifiers, we outperform previous approaches on three pedestrian datasets, INRIA, ETH, and Caltech USA. We present a new approach to train such classifiers. By reusing computations among different training stages, 16 occlusion-specific classifiers can be trained at only one tenth the cost of one full training. We show that also test time cost grows sub-linearly.Keywords
This publication has 12 references indexed in Scilit:
- Seeking the Strongest Rigid Detector2013
- Pedestrian detection at 100 frames per second2012
- Detection and Tracking of Occluded People2012
- Pedestrian Detection: An Evaluation of the State of the ArtIEEE Transactions on Pattern Analysis and Machine Intelligence, 2011
- Monocular 3D scene understanding with explicit occlusion reasoning2011
- An HOG-LBP human detector with partial occlusion handling2009
- Integral Channel Features2009
- Detection of multiple, partially occluded humans in a single image by Bayesian combination of edgelet part detectors2005
- Sharing features: efficient boosting procedures for multiclass object detection2004
- Robust Real-Time Face DetectionInternational Journal of Computer Vision, 2004