R. Vaillant, C. Monrocq, and Y. LeCun. Original approach for the localisation of objects in images. IEE Proc on Vision, Image, and Signal Processing, 141(4):245-250, August 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....early face detection systems eschewed component based architectures for a holistic approach. In [14] the distribution of faces is modelled with a mixture of gaussian curves. Faces are detected by measuring comparing novel patterns to the model distribution. A similar approach is taken in [12] and [15, 19], where a single SVM and a set of neural networks, respectively, are trained to discriminate between face and non face patterns. In [14, 15] virtual examples, as explored in [11] are incorporated by rotating, translating, or scaling positive face examples, and including these new training points ....

R. Vaillant, C. Monrocq, and Y. LeCun. An original approach for the localisation of objects in images. In International Conference on Artificial Neural Networks, pages 26--30, 1993.

....classification than PCA. The patterns in each cluster are supposed to have Gaussian distribution. The maximum likelihood decision rule for discriminating between face and non pattern is used in both methods. Several methods that use neural networks to localize the human face have been proposed [47, 48, 1, 53, 33]. The most significant work is probably by Rowley et al. 38] Their system uses several multilayer neural networks and a simple arbitration scheme like ANDing, ORing and voting to improve performance over a single network. One of the limitations of this system is that only frontal and upright ....

R. Vaillant, C. Monrocq, and Y. L. Cun. An original approach for the localisation of objects in images. In Proc. of Vision, Image and Signal Processing, pages 3:57--60, 1994.

....classifier were successfully applied to tasks where the pose of the object was fixed. In [6] Haar wavelet features are used to detect frontal and back views of pedestrians with an SVM classifier. Learning based systems for detecting frontal faces based on a gray value features are described in [14, 13, 10, 2]. Component based techniques promise to provide more invariance since the individual components vary less under pose changes than the whole object. Variations induced by pose changes occur mainly in the locations of the components. A component based method for detecting faces based on the ....

R. Vaillant, C. Monrocq, and Y. Le Cun. An original approach for the localisation of objects in images. In International Conference on Artificial Neural Networks, pages 26--30, 1993.

....single template was successfully applied to tasks where the pose of the object was fixed. In [5] Haar wavelet features were used to detect frontal and back views of pedestrians with a SVM classifier. Learning based system for detecting frontal faces based on a gray value features are described in [14, 12, 9, 2]. Component based techniques promise to provide more invariance since the individual components vary less while the variations induced by pose changes occur mainly in their geometry. A component based method for detecting faces based on the empirical probabilities of overlapping rectangular parts ....

R. Vaillant, C. Monrocq, and Y. Le Cun. An original approach for the localisation of objects in images. In International Conference on Artificial Neural Networks, pages 26--30, 1993.

.... type are image invariance methods which base a matching on a set of image pattern relationships (e.g. brightness levels) that, ideally, uniquely determine the objects being searched for [21] The final set of object detection systems are characterized by their example based learning algorithms [24], 22] 23] 18] 19] 16] 14] These systems learn the salient IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. 23, NO. 4, APRIL 2001 349 . A. Mohan and C. Papageorgiou are with Kana Communications, 740 Bay Road, Redwood City, CA 94063. E mail: amohan alum.mit.edu, ....

R. Vaillant, C. Monrocq, and Y. Le Cun, Original Approach for the Localisation of Objects in Images, IEE Proc. Vision Image Signal Processing, vol. 141, no. 4, pp. 245-50, Aug. 1994.

.... fields and maximum a posteriori probability estimation is developed in [7] Kullback relative information is used for maximal discrimination between positive and negative examples of faces whose densities are modeled by discrete Markov processes in [8] Other techniques include neural networks [9], or algorithms where feature points are detected using spatial filters and then grouped into face candidates using geometric and gray level constrains [10] Fast face detection using multilayer perceptrons and fast Fourier transform is described in [11] For the detection of upright, frontal This ....

R. Vaillant, C. Monrocq, and Y. Len Cun, "Original approach for the localisation of objects in images ," IEE Proc. Vis. Image Signal Processing, vol. 141, no. 4, August 1994.

....as people, since they involve a significant amount of prior information and domain knowledge. In recent research the detection problem has been solved using learning based techniques that are data driven. This approach was used by Sung and Poggio [ Sung and Poggio, 1994 ] and Vaillant, et al. Vaillant et al. 1994 ] for the detection of frontal faces in cluttered scenes, with similar architectures later used by Moghaddam and Pentland [ Moghaddam and Pentland, 1995 ] Rowley, et al. Rowley et al. 1995 ] and Osuna et al. Osuna et al. 1997b ] The image representations used were either projections ....

R. Vaillant, C. Monrocq, and Y. Le Cun. Original approach for the localisation of objects in images. IEEE Proc. Vis. Image Signal Process., 141(4), August 1994.

....in video sequences focusing on using motion and 3D models or constraints to find people [60, 31, 25, 49, 70, 23, 34] In recent research the detection problem has been solved using learning based techniques that are data driven. This approach was used by Sung and Poggio[56] and Vaillant, et al. [62] for the detection of frontal faces in cluttered scenes, with similar architectures later used by Moghaddam and Pentland [36] Rowley, et al. 51] and Osuna et al. 40] We now briefly discuss how the learning mechanisms outlined in the first part of the paper can be used as an approach to object ....

R. Vaillant, C. Monrocq, and Y. Le Cun. Original approach for the localisation of objects in images. IEE Proceedings Vision Image Signal Processing, 141(4):245--50, August 1994.

....more complex objects such as people, since they involve a significant amount of prior information and domain knowledge. In recent research the detection problem has been solved using learning based techniques that are data driven. This approach was used by Sung and Poggio[17] and Vaillant, et al. [19] for the detection of frontal faces in cluttered scenes, with similar architectures later used by Moghaddam and Pentland [10] Rowley, et al. 16] and Osuna et al. 12] A variety of image representations have been used in these works. In this paper we compare some of them that we also link through ....

R. Vaillant, C. Monrocq, and Y. Le Cun. Original approach for the localisation of objects in images. IEE Proceedings Vision Image Signal Processing, 141(4):245--50, August 1994.

.... 11 92.9 64 1 in 151220 [Sung and Poggio, 1994] Multi layer network) 36 76.8 5 1 in 1929655 [Sung and Poggio, 1994] Perceptron) 28 81.9 13 1 in 742175 The candidate verification process used to speed up our system, described in Section 4, is similar to the detection technique used by [V aillant et al. 1994] In that work, two networks were used. The first network has a single output, and like our system it is trained to produce a maximal positive value for centered faces, and a maximal negative value for non faces. Unlike our system, for faces that are not perfectly centered, the network is ....

R. Vaillant, C. Monrocq, and Y. Le Cun. Original approach for the localisation of objects in images. IEE Proceedings on Vision, Image, and Signal Processing, 141(4), August 1994.

....centered faces. We use the centered face detector to verify candidates found by the translation invariant network. With this approach, we can process a 320x240 pixel image in less than 5 seconds on an SGI Indy workstation. This technique is related, at a high level, to the technique presented in [V aillant et al. 1994] 5 Conclusions and future research Our algorithm can detect between 78.9 and 90.5 of faces in a set of 130 test images, with an acceptable number of false detections. Depending on the application, the system can be made more or less conservative by varying the arbitration heuristics or ....

R. Vaillant, C. Monrocq, and Y. Le Cun. Original approach for the localisation of objects in images. IEE Proceedings on Vision, Image, and Signal Processing, 141(4), August 1994.

.... searched for (Sinha, 1994[22] 23] More recently, systems for detecting unoccluded vertical frontal views of human faces in images have been developed using example based approaches by Sung and Poggio, 1994[26] Moghaddam and Pentland, 1995[15] Rowley et al. 1995[20] Vaillant et al. 1994[28], and Osuna et al. 1997[17] These view based approaches can handle detecting faces in cluttered scenes and have shown a reasonable degree of success when extended to handle non frontal views. The system of Sung and Poggio models a set of faces and non faces as clusters in a high dimensional ....

R. Vaillant, C. Monrocq, and Y. Le Cun. Original approach for the localisation of objects in images. IEE Proc.-Vis. Image Signal Processing, 141(4), August 1994.

....empirical results are also provided. 1 Introduction This paper presents a general method to extend many template based frontal, upright, face detection systems to handle in plane rotations of the face. There have been many template based face detection systems developed, for example, see [2, 3, 6 8, 11 13, 15]. Other systems, such as [5] can also achieve rotation invariance by extracting smaller features of the face and using graph matching algorithms. In this paper, we concentrate on template based methods, in particular the one presented by Rowley, Baluja Kanade in [11] The simplest method for ....

R. Vaillant, C. Monrocq, and Y. Le Cun. Original approach for the localisation of objects in images. IEE Proceedings on Vision, Image, and Signal Processing, 141(4), August 1994.

....at many possible scales and uses a SVM as its core classification algorithms to determine the appropriate class (face non face) 3.1 Previous Systems The problem of face detection has been approached with different techniques in the last few years. This techniques include Neural Networks [2, 9, 11], detection of face features and use of geometrical constraints [13] density estimation of the training data [6] labeled graphs [5] and clustering and distribution based modeling [10] Out of all these previous works, the results of Sung and Poggio [10] and Rowley et al. 9] reflect systems ....

R. Vaillant, C. Monrocq, and Y. Le Cun. Original approach for the localisation of objects in images. IEEE Proc. Vis. Image Signal Process., 141(4), August 1994.

....scenes is the face detection system of Sung and Poggio [ Sung and Poggio 1994 ] They model face and non face patterns in a high dimensional space and derive a statistical model for the class of frontal human faces. Similar face detection systems have been developed by others (Vaillant, et al. Vaillant et al. 1994 ] Rowley, et al. Rowley et al. 1995 ] Moghaddam and A. Pentland [ Moghaddam and Pentland 1995 ] Osuna et al. Edgar Osuna and Girosi 1996 ] Frontal human faces, despite their variability, share very similar patterns (shape and the spatial layout of facial features) and their color ....

R. Vaillant, C. Monrocq, and Y. Le Cun. Original approach for the localisation of objects in images. IEE Proc.-Vis. Image Signal Processing, 141(4), August 1994.

....for faces. Many face detection researchers have used the idea that facial images can be characterized directly in terms of pixel intensities. These images can be characterized by probabilistic models of the set of face images [4, 13, 15] or implicitly by neural networks or other mechanisms [3, 12, 14, 19,21,23,25,26]. The parameters for these models are adjusted either automatically from example images (as in our work) or by hand. A few authors have taken the approach of extracting features and applying either manually or automatically generated rules for evaluating these features [7,11] Training a neural ....

....range of face types under good lighting with uncluttered backgrounds, while Test Set 1 tests the robustness to variable lighting and cluttered backgrounds. The candidate verification process used to speed up our system, described in Section 4, is similar to the detection technique presented in [23]. In that work, two networks were used. The first network has a single output, and like our system it is trained to produce a positive value for centered faces, and a negative value for nonfaces. Unlike our system, for faces that are not perfectly centered, the network is trained to produce an ....

R. Vaillant, C. Monrocq, and Y. Le Cun. Original approach for the localisation of objects in images. IEE Proceedings on Vision, Image, and Signal Processing, 141(4), August 1994.

....to detect faces in gray scale images. Unlike similar previous systems which could only detect upright, frontal faces [Sung, 1996, Rowley et al. 1998, Moghaddam and Pentland, 1995, Pentland et al. 1994, Burel and Carel, 1994, Colmenarez and Huang, 1997, Osuna et al. 1997, Lin et al. 1997, Vaillant et al. 1994, Yang and Huang, 1994, Yow and Cipolla, 1996] this system efficiently detects frontal faces which can be arbitrarily rotated within the image plane. We also present preliminary results on detecting upright faces which are rotated out of the image plane, such as profiles and semi profiles. Many ....

.... These images can be characterized by probabilistic models of the set of face images [Colmenarez and Huang, 1997, Moghaddam and Pentland, 1995, Pentland et al. 1994] or implicitly by neural networks or other mechanisms [Burel and Carel, 1994, Osuna et al. 1997, Rowley et al. 1998, Sung, 1996, Vaillant et al. 1994, Yang and Huang, 1994] Other researchers have taken the approach of extracting features and applying either manually or automatically generated rules for evaluating these features. By using a graph matching algorithm on detected features, Leung et al. 1995] can also achieve rotation ....

R. Vaillant, C. Monrocq, and Y. Le Cun. Original approach for the localisation of objects in images. IEE Proceedings on Vision, Image, and Signal Processing, 141(4), August 1994.

....of face detector demonstrations, we have found that users expect faces to be detected at any angle, as shown in Figure 1. In this paper, we present a neural network based algorithm to detect faces in gray scale images. Unlike similar previous systems which could only detect upright, frontal faces [3, 4, 6 9, 12, 13, 15, 17, 18], this system efficiently detects frontal faces which can be arbitrarily rotated within the image plane. We also present preliminary results on detecting upright faces rotated out of the image plane, such as profiles and semi profiles. Many face detection systems are template based; they encode ....

....such as profiles and semi profiles. Many face detection systems are template based; they encode facial images directly in terms of pixel intensities. These images can be characterized by probabilistic models of the set of face images [4, 7, 9] or implicitly by neural networks or other mechanisms [3, 6, 8, 12,13,15, 17]. Other researchers have taken the approach of extracting features and applying either manually or automatically generated rules for evaluating these features [5,18] By using a graphmatching algorithm on detected features, 5] also demonstrated rotation invariance. We present a general method to ....

R. Vaillant, C. Monrocq, and Y. Le Cun. Original approach for the localisation of objects in images. IEE Proceedingson Vision, Image, and Signal Processing, 141(4), August 1994.

....the object of interest in the corresponding receptive field. Since the size of the objects to be detected within the image are unknown, the image can be presented to the network at multiple resolutions, and the results at multiple resolutions combined. The idea has been applied to face location, [Vaillant, Monrocq and LeCun 1994], address block location on envelopes [Wolf and Platt 1994] and hand tracking in video [Nowlan and Platt 1995] To illustrate the method, we will consider the case of face detection in images as described in [Vaillant, Monrocq and LeCun 1994] First, images containing faces at various scales are ....

....combined. The idea has been applied to face location, Vaillant, Monrocq and LeCun 1994] address block location on envelopes [Wolf and Platt 1994] and hand tracking in video [Nowlan and Platt 1995] To illustrate the method, we will consider the case of face detection in images as described in [Vaillant, Monrocq and LeCun 1994]. First, images containing faces at various scales are collected. Those images are filtered through a zero mean Laplacian filter so as to remove variations in global illumination and large scale illumination gradients. Then, training samples of faces and non faces are manually extracted from these ....

Vaillant, R., Monrocq, C., and LeCun, Y. (1994). Original approach for the localisation of objects in images. IEE Proc on Vision, Image, and Signal Processing, 141(4):245--250.

....respect to translations, scaling, skewing, and other distortions. They can directly accept images with no preprocessing other than a simple size normalization and centering. They have had numerous applications in handwriting recognition [10, 2, 9] and object location in images, particulaly faces [13]. The architecture of LeNet5 is shown in figure 5. In a convolutional net, each unit takes its input from a local receptive field on the layer below, forcing it to extract a local feature. Furthermore, units located at different places on the image are grouped in planes, called feature maps, ....

R. Vaillant, C. Monrocq, and Y. Le Cun. Original approach for the localisation of objects in images. IEE Proc on Vision, Image, and Signal Processing, 141(4), August 1994.

No context found.

R. Vaillant, C. Monrocq, and Y. LeCun. Original approach for the localisation of objects in images. IEE Proc on Vision, Image, and Signal Processing, 141(4):245-250, August 1994.

No context found.

R. Vaillant, C. Monrocq, and Y. LeCun. An original approach for the localisation of objects in images. In International Conference on Artificial Neural Networks, pages 26--30, 1993.

No context found.

R. Vaillant, C. Monrocq, and Y. LeCun. An original approach for the localisation of objects in images. In International Conference on Artificial Neural Networks, pages 26--30, 1993. Proceedings of the IEEE International Workshop on Analysis and Modeling of Faces and Gestures (AMFG'03)

No context found.

R. Vaillant, C. Monrocq, and Y. Le Cun. Original approach for the localisation of objects in images. IEEE Proc. Vis. Image Signal Process., 141(4), August 1994.

First 50 documents

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