T. D. Rikert, M. J. Jones and P. Viola, A Cluster-Based Statistical Model for Object Detection, Proceedings of the International Conference on Computer Vision-Volume 2 - Volume 2 Page: 1046 Year of Publication: 1999  Home/Search   Document Details and Download   Summary   Related Articles   Check

....for each filter) and thirdly, simply more filters were used than before typically between ten and fifty filters or wavelets to measure texture fea tures at a set of scales and orientations. These filter response distributions were learnt from training images and represented by clusters [2, 11, 15, 17], or histograms [9, 10, 19] The distributions could then be used for classification, segmentation or synthesis. For instance, classification could be achieved by comparing the distribution of a novel texture image to the model distributions learnt from the texture classes. Similarly, synthesis ....

T.D. Rikert, M. J. Jones, and P. A. Viola. A cluster-based statistical model for object detection. In Proc. ICCV, 1999.

....of faces is proposed in [4] It uses local feature extractors to detect the eyes, the corner of the mouth, and the tip of the nose. The geometrical configuration of these features is matched with a model configuration by conditional search. A related method using statistical models is published in [9]. Local features are extracted by applying multi scale and multi orientation filters to the input image. The responses of the filters on the training set are modeled as Gaussian distributions. In [5] pedestrian detection is performed by a set of SVM classifiers each of which was trained to detect ....

T. D. Rikert, M. J. Jones, and P. Viola. A cluster-based statistical model for object detection. In Proc. IEEE Conference on Computer Vision and Pattern Recognition, volume 2, pages 1046--1053, Fort Collins, 1999.

....Statistical models for detection of objects and their attributes in images have been popular for a number of years. Different models of varying degrees of complexity have been proposed and utilized in tasks such as edge and color detection [1, 2, 3, 4, 5] and face detection and recognition[6, 7, 8, 9], to name just a few. A common thread of these approaches is that their improved performance usually came at the cost of high computational and algorithmic complexity. Similarly, computationally simpler techniques often required large amounts of data to compensate for the weaknesses of the model. ....

....face detection is a difficult task and various methods that have been used are often complex, we show how a simple boosted algorithm can be used to accurately 3 and efficiently solve this detection task. A number of models, primarily statistical, have explored various ways of modeling faces [6, 7, 8, 9]. Unlike the task of color based segmentation which imposes little if any spatial constraints on the model, detection of faces also relies on spatial constraints. To extend the proposed boosting framework to detection of faces we consider a set of simple weak classifiers of the following form: ....

T. Rikert, M. Jones, and P. Viola, "A cluster-based statistical model for object detection," in ICCV, 1999.

....as for example faces or zebras. More recent methods construct models and localize them in the image. They differ in the model representation and in the learning algorithm. Models are for example represented by global images patches [17] geometric relations of parts [18] or statistical models [14]. Learning algorithms are either supervised or unsupervised. Supervised algorithms require the manual extraction of regions or features. In the unsupervised case images are labeled as positive or negative which avoids time consuming manual intervention. In this paper we propose an unsupervised ....

....Positions of each attribute are represented with respect to a coordinate frame fixed to the object. Their representation is rigid, but allows for small positional variations. It is learnt from a large set of manually extracted examples. Characteristic distributions of feature vectors are learnt by [8, 14]. These methods require the manual extraction and annotation of regions. They can represent for example faces, road or sky. Amit and Geman [1] learn a hierarchical model from edge features. They select distinctive local feature groupings of edgels constrained by loose geometrical relationships and ....

[Article contains additional citation context not shown here]

T.D. Rikert, M.J. Jones, and P. Viola. A cluster-based statistical model for object detection. In ICCV, pp. 1046--1053, 1999.

....of faces, is proposed in [3] by using local feature extractors to detect the eyes, the corner of the mouth and tip of the nose. The geometrical configuration of these features is matched with a model configuration by conditional search. A related method using statistical models is published in [8]. Local features are extracted by applying multi scale and multi orientation filters to the input image. The responses of the filters on the training set are modeled as Gaussian distributions. In [4] pedestrian detection was performed by a set of SVM classifiers each of which was trained to detect ....

T. D. Rikert, M. J. Jones, and P. Viola. A cluster-based statistical model for object detection. In Proc. IEEE Conference on Computer Vision and Pattern Recognition, volume 2, pages 1046-- 1053, Fort Collins, 1999.

....the eyes, corner of the mouth, and tip of the nose. Assuming that the position of the eyes is properly determined, the geometrical con guration of the detected parts in the image is matched with a model con guration by conditional search. A related method using statistical models is published in [Rikert et al. 99] Local features are extracted by applying multi scale and multi orientation lters to the input image. The responses of the lters on the training set are modeled as Gaussian distributions. In contrast to [Leung et al. 95] the con guration of the local lter responses is not matched with a ....

T. D. Rikert, M. J. Jones, P. Viola. A cluster-based statistical model for object detection. Proc. IEEE Conference on Computer Vision and Pattern Recognition, 1999, 1046-1053.

....is rigid. Furthermore, a global representation requires learning in the presence of high dimensional data which is a difficult problem. It is also more difficult to make a global representation invariant to image transformations. More recent approaches therefore use a local face representation [2, 7, 10]. Burl et al. 2] detect facial features by correlation and use spatial constraints between these features for verification. Their face representation is not learnt but selected manually. This is the main different with our approach where facial features as well as spatial constraints are learnt ....

....are learnt from a set of samples. Wu et al. 10] build two fuzzy models to describe skin color and hair color, respectively. These models are used to extract regions in an unknown image. These regions are then compared with a head shaped model using a fuzzy pattern matching method. Rikert et al. [7] describe a face image by a set of local feature vectors (Gabor filters) Feature vectors of the training images are clustered and the most discriminant clusters are used to describe the face class. They don t use any spatial constraints for verification. 1.2 Overview of the paper Section 2 ....

T. Rikert, M. Jones, and P. Viola. A cluster-based statistical model for object detection. In ICCV, 1999.

No context found.

T. D. Rikert, M. J. Jones and P. Viola, A Cluster-Based Statistical Model for Object Detection, Proceedings of the International Conference on Computer Vision-Volume 2 - Volume 2 Page: 1046 Year of Publication: 1999

No context found.

Thomas D. Rikert, Michael J. Jones, and Paul Viola. A Cluster-Based Statistical Model for Object Detection. ICCV 1999.

No context found.

T. Rikert, M. Jones, and P. Viola. A cluster-based statistical model for object detection. In Proceedings International Conference On Computer Vision, 1999.

No context found.

T.D. Rikert, M.J. Jones, and P. Viola. A cluster-based statistical model for object detection. In ICCV, pages 1046--1053, 1999.

No context found.

T. Rikert, M. Jones, P. Viola, A cluster-based statistical model for object detection, Proceedings of the IEEE International Conference on Computer Vision (1999) 1046 -- 1053.

No context found.

T. D. Rikert, M. J. Jones, P.Viola.A cluster-based statistical model for object detection. Proc. IEEE Conference on Computer Vision and Pattern Recognition, 1999, 1046-1053.

No context found.

T. D. Rikert, M. J. Jones, and P. Viola. A cluster-based statistical model for object detection. In Proc. IEEE Conference on Computer Vision and Pattern Recognition, volume 2, pages 1046--1053, Fort Collins, 1999.

No context found.

T. D. Rikert, M. J. Jones, and P. Viola, "A cluster-based statistical model for object detection," Proc. IEEE Conf. Computer Vision and Pattern Recognition, 1999.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC