Multi-class object detection with hough forests using local histograms of visual words

Authors:
Markus Mühling;Ralph Ewerth;Bing Shi;Bernd Freisleben
Affiliations:
Department of Mathematics & Computer Science, University of Marburg, Marburg, Germany;Department of Mathematics & Computer Science, University of Marburg, Marburg, Germany;Department of Mathematics & Computer Science, University of Marburg, Marburg, Germany;Department of Mathematics & Computer Science, University of Marburg, Marburg, Germany
Venue:
CAIP'11 Proceedings of the 14th international conference on Computer analysis of images and patterns - Volume Part I
Year:
2011

Citing 9
Cited 0

Random Forests

Machine Learning
Distinctive Image Features from Scale-Invariant Keypoints

International Journal of Computer Vision
Learning generative visual models from few training examples: An incremental Bayesian approach tested on 101 object categories

Computer Vision and Image Understanding
Sharing Visual Features for Multiclass and Multiview Object Detection

IEEE Transactions on Pattern Analysis and Machine Intelligence
Towards optimal bag-of-features for object categorization and semantic video retrieval

Proceedings of the 6th ACM international conference on Image and video retrieval
Robust Object Detection with Interleaved Categorization and Segmentation

International Journal of Computer Vision
Improving Semantic Video Retrieval via Object-Based Features

ICSC '09 Proceedings of the 2009 IEEE International Conference on Semantic Computing
The Pascal Visual Object Classes (VOC) Challenge

International Journal of Computer Vision
Vlfeat: an open and portable library of computer vision algorithms

Proceedings of the international conference on Multimedia

Quantified Score

Hi-index	0.00

Visualization

Abstract

Multi-class object detection is a promising approach for reducing the processing time of object recognition tasks. Recently, random Hough forests have been successfully used for single-class object detection. In this paper, we present an extension of random Hough forests for the purpose of multi-class object detection and propose local histograms of visual words as appropriate features. Experimental results for the Caltech-101 test set demonstrate that the performance of the proposed approach is almost as good as the performance of a single-class object detector, even when detecting a large number of 24 object classes at a time.