Analysis and optimization of parameters used in training a cascade classifier
DOI:
https://doi.org/10.14738/aivp.32.1152Keywords:
Cascade classifier, Local Binary Pattern (LBP), Histogram of Gradients (HOG), Object detection, face detection, OpenCVAbstract
Training a cascade classifier for object detection using Local Binary Pattern (LBP) and Histogram of Gradients (HOG) features is computationally exorbitant. If the parameters of training are not chosen appropriately, the training may take weeks to complete with the output of an inefficient classifier. The state-of-the-art face recognition applications demand accurate and reliable cascade classifiers. Open Computer Vision (OpenCV) organization provides libraries which accomplish the training task once all parameters are given as inputs. In this paper we analyze the parameters experimentally and concluded with an optimal range of values for each of these parameters. Testing of the generated classifiers with optimal parameters values is performed on a dataset of 4000 test images. The training of these classifiers with optimal parameters takes an average training time of 25000 sec and provides average true positive detection of 88%.References
Muller, N., Magaia, L. and Herbst B.M. Singular value decomposition, eigenfaces, and 3D reconstructions. SIAM Review, 2004, 46(3): p. 518-545.
Bradski, G. Computer vision face tracking for use in a perceptual user interface, Intel Technology Journal, 2nd Quarter, 1998.
Viola, P. and Jones, M. Rapid object detection using boosted cascade of simple features, IEEE Conference on Computer Vision and Pattern Recognition, 2001, 1: p.511-518.
Pentland, A, Looking at People: Sensing for ubiquitous and wearable computing, IEEE trans. on Pattern Analysis and Machine Intelligence, 2000, 22(1): p. 107-119.
Li, S. Z., Zhu, L., Zhang, Z., Blake, A., Zhang, H., and Shum, H., Statistical learning of multi-view face detection, In 7th European Conference Computer Vision, 2002, part iv: p. 67–81.
Schneiderman, H. and Kanade, T, A statistical method for 3d object detection applied to faces and cars, In IEEE Conference on Computer Vision and Pattern Recognition, 2000, vol 1: p. 1746–1759.
Viola, P. and Jones, M. J., Robust real-time face detection, International Journal of Computer Vision, 2004, 57(2): p.151–173.
Lienhart, R. and Maydt, J., An extended set of haarlike features for rapid object detection, In IEEE International conference on Image processing, 2002, vol 1, p. 900–903.
Y. Freund and R. E. Schapire. Experiments with a new boosting algorithm, In proc. of International conference on Machine Learning, 1996, p.148-156.
M.-T. Pham and T.-J. Cham, Online learning asymmetric boosted classifiers for object detection, In proc. of Computer vision and pattern recognition, 2007, p. 1-8.
P. Viola and M. Jones., Fast and robust classification using asymmetric adaboost and a detector cascade, In Neural information processing systems, MIT Press, Cambridge, MA, 2002.
R. Lienhart, A. Kuranov, and V. Pisarevsky., Empirical analysis of detection cascades of boosted classifiers for rapid object detection, In DAGM-Symposium, 2003, p. 297–304.
C. Huang, H. Ai, Y. Li, and S. Lao, High-performance rotation invariant multiview face detection, in IEEE Transactions on Pattern Analysis and Machine Intelligence, TPAMI, 2007, 29:671–686.
B. Wu, H. Ai, C. Huang, and S. Lao. Fast rotation invariant multi-view face detection based on real adaboost. In proc. of International Conference on Automatic face and Gesture recognition, 2004, p. 79–84.
Dataset-1, Caltech 101 Object categories, (available online), http://www.vision.caltech.edu/Image_Datasets/Caltech101/
Dataset-2, Amos Storkey, Image sequence segmentation data, (available online) http://homepages.inf.ed.ac.uk/amos/afreightdata.html
Dataset-3, Oriental Bird Images, (Available online), http://orientalbirdimages.org/
Dataset-4, Computational Visual Cognition Laboratory, MIT, (Available online), http://cvcl.mit.edu/database.htm
Jerome H. Friedman, Greedy function approximation: A gradient boosting machine, IMS 1999 Reitz lecture (Available online), https://statweb.stanford.edu/~jhf/ftp/trebst.pdf
