Fleck, M. M., Forsyth, D. A., and Bregler, C. Finding Naked People. ECCV, 1996, pages 593-602.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....color histograms [3, 18] parametric flow models [21] models of facial dynamics [4, 7] Hidden Markov Models [14] and stereo systems [12] We are unaware of any previous work that ad dressed sclera detection. Skin color based detection of faces, however, has been explored extensively, e.g. [8, 16, 20, 22]. Gaze estimation has proven to be a challenging problem. Previous approaches include systems based on neural networks [1, 17] morphable models [15] and self organizing gray scale units [2] Gee and Cipolla [9] explore the underlying geometric constraints. Statistical Sclera and Skin Color ....

M. Fleck, D. Forsyth, and C. Bregler. Finding naked people. In Lecture Notes in Computer Science. Vol. 1065: Proceedings of the jth European Conference on Computer Vision, volume II, pages 592-602. Springer-Verlag, Berlin, April 1996.

....to detect occluded people or people whose body parts have little contrast with the background. The system needs many transformations and match with wavelet template, and it mainly deals with pedestrian detection. David A. Forsyth et al. present some methods to find naked people in a static image [For96a, For97a, For99a, Iof01a]. The skin of human has some properties in hue and saturation. They segment the regions with these properties, and detect edges in each region. They get rectangles by combining the edges based on local symmetry. At last, they use a complex probability function to decide which combination of ....

D.Forsyth and M.Fleck. Finding naked people. In: Proceedings of European Conference of Computer Vision. Berlin, Germany: SpringerVerlag, 593-602, 1996.

....and their faces and eyes in real time. Facial motions have been analyzed in real time or near real time using parametric flow models [18] models of facial dynamics [5] Hidden Markov Models [12] and stereo systems [9] Skin color based detection of faces has been explored extensively, e.g. [7, 17]. Gaze estimation in ambient light has been addressed with neural networks [1, 16] morphable models [13] and self organizing gray scale units [2] Gaze estima tion in infrared light has been explored extensively, for example, to analyze driver behavior [14] Video analysis to detect ....

M. Fleck, D. Foyth, and C. Bregler. Finding naked people. In Lecture Notes in Computer Science. Vol. 1065: Proceedings of the Jth European Conference on Computer Vision, volume II, pages 592-602. Springer-Verlag, Berlin, April 1996.

....about the objects to simplify the segmentation process. Forsyth and Fleck [10] describe a representation for animals as an assembly of almost cylindrical parts. On a database of images of animals, their representation can retrieve images of horses, for example, in a variety of poses. Fleck et al. [8] use knowledge about the positions of attachment of limbs and head to the human body to detect the presence of naked people in the database images. Forsyth et al. illustrate some specialized applications of image retrieval in [9] There has been a lot of work in the area of image segmentation. ....

M. M. Fleck, D. A. Forsyth, and C. Bregler. Finding naked people. Fourth European Conference on Computer Vision, pages 72-77, 1996.

....and their faces and eyes in real time. Facial motions have been analyzed in real time or near real time using parametric ow models [18] models of facial dynamics [5] Hidden Markov Models [12] and stereo systems [9] Skin color based detection of faces has been explored extensively, e.g. [7, 17]. Gaze estimation in ambient light has been addressed with neural networks [1, 16] morphable models [13] and self organizing gray scale units [2] Gaze estimation in infrared light has been explored extensively, for example, to analyze driver behavior [14] Video analysis to detect driver ....

M. Fleck, D. Forsyth, and C. Bregler. Finding naked people. In Lecture Notes in Computer Science. Vol. 1065: Proceedings of the 4th European Conference on Computer Vision, volume II, pages 592-602. Springer-Verlag, Berlin, April 1996.

....wish to block them. Our approach is designed to support a wide range of organisational policies on the circulation of images containing large areas of skin and sexually related vocabulary. 2 Image based analysis Algorithms to identify skin form a common module in many computer vision systems ([6, 7, 8] for example) This Section compares these algorithms and illustrates some possible high level features that might be useful for classifying 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 g b Figure 1: Left: original skin with skin region highlighted; ....

....Our first objective was to choose a colour space in which the skin region was as compact as possible. Each pixel, #####in the training set is transformed to one of the colour spaces shown in Table 1. Colour space Components RGB ##### HSV ## ## # Normalised RGB Two o f # ## # ## Log opponent [8] ### # ## # Comprehensive [9] Two o f # ## # ## Table 1: Colour space conventions. For the normalised RGB and the comprehensive normalisation intensity variation is removed so one colour component is a linear combination of the other two. The HSV colour space [10] may be derived from the RGB ....

[Article contains additional citation context not shown here]

Margaret M. Fleck, David A. Forsyth, and Chris Bregler. Finding naked people. In European Conference on Computer Vision, volume II, pages 593--602. Springer-Verlag, 1996.

....vocabulary. 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0. 9 Figure 1: Left: original skin with skin region highlighted; Right: RGB plot of skin pixels 2 Image based analysis Algorithms to identify skin form a common module in many computer vision systems ([6, 7, 8] for example) This Section compares these algorithms and illustrates some possible high level features that might be useful for classifying images containing people. 2.1 Skin filtering A number of competing approaches have been proposed for the identification of pixels that are skin coloured. ....

.... The HSV colour space [10] may be derived from the RGB space as v =max(r, g, b) s=d v, h = 8 : g b r = v 2 r b g = v 4 g r b = v (1) Figure 2: Example clothed images from the training set Colour space Components RGB r, g, b HSV h, s, v Normalised RGB Two o f r, g, Log opponent [8] I,R g ,B y Comprehensive [9] Two o f r, g, b Table 1: Colour space conventions. For the normalised RGB and the comprehensive normalisation intensity variation is removed so one colour component is a linear combination of the other two. where d =max(r, g, b) min(r, g, b) The log opponent ....

[Article contains additional citation context not shown here]

Margaret M. Fleck, David A. Forsyth, and Chris Bregler. Finding naked people. In European Conference on Computer Vision, volume II, pages 593--602. Springer-Verlag, 1996.

....as acceptable or inappropriate. The problem therefore is the non retrieval of certain types of image. Although the identification of human skin is commonplace in vision systems, the detection of pictures containing nudity and pornography is a fairly specialised area (some relevant systems include [1 3] and [4, 5] These systems contain a skin filter which is usually based on colour sometimes with texture as a secondary feature. Skin filters are now fairly standard so we give only a brief explanation Section 2. Here we wish to focus on the classification and deployment of such systems which we ....

....proportions of images were chosen to be broadly representative of a range of commercial environments but we know there is considerable variation in these priors between sites. This issue in discussed further later. There are suggestions for high level features based on grouping of skin segments [1] that might distinguish these classes but here we have a requirement to process the images speedily so, along with [2] and [3] are interested to try simpler features. For each blob in the image we have computed: area; centroid; the length of the major axis of an ellipse with the same second order ....

[Article contains additional citation context not shown here]

Fleck, M.M., Forsyth, D.A., Bregler, C.: Finding naked people. In: European Conference on Computer Vision. Volume II., Springer-Verlag (1996) 593--602

....using specific examples. After ensuring that these components are present in the proper geometric configuration, a second example based classifier combines the results of the component detectors to classify a pattern as either a person or a non person. A similar part based approach is followed in [24] to detect naked people. First, large skin colored components are found in an image by applying a skin filter that combines color and texture. Based on geometrical constraints between detected components an image is labeled as containing naked people or not. Obviously this method is suited for ....

M.M. Fleck, D.A. Forsyth, and C. Bregler. Finding naked people. In European Conference on Computer Vision, volume 2, pages 593--602, Cambridge, UK, 1996.

....skin or non skin label. Oliver et al. 13] used spatial coordinates in addition to the normalized RG to obtain blobs as a low level image feature. By estimating mean vector and covariance matrix for 2D and 3D blobs, Gaussian mixture models of skin data were obtained in the colorspace. Fleck et al. [7] presented a skin detection method to report if images contain the naked people using skin filter. The log transformed colorspaces from RGB have been used. The skin filter is created through log opponent based color representation and texture amplitude in green channel to distinguish between skin ....

M.M. Fleck, D.A. Forsyth and C. Bregler, "Finding naked People," ECCV, Vol II, pp. 592-602. 1996.

....using speci c examples. After ensuring that these components are present in the proper geometric con guration, a second example based classi er combines the results of the component detectors to classify a pattern as either a person or a non person. A similar part based approach is followed in [24] to detect naked people. First, large skin colored components are found in an image by applying a skin lter that combines 10 color and texture. Based on geometrical constraints between detected components an image is labeled as containing naked people or not. Obviously this method is suited for ....

M.M. Fleck, D.A. Forsyth, and C. Bregler. Finding naked people. In European Conference on Computer Vision, volume 2, pages 593-602, Cambridge, UK, 1996.

....suitable for all colour images does not exist. Thus it is not surprising that several different colour spaces have been investigated for skin classification within the colour based approaches. A number of researchers have shown that skin colours fall in to a very narrow band in the colour space [14, 143,157, 158]. This is illustrated in Figures 9.1, 9.2 and 9.3, taken from [14] The plots show skin ratios for R G, RIB and G B respectively, where R, G and B are the corresponding values of red, blue and green. Two plots can be seen for each ratio: the first uses hand labelled skin data; the second uses ....

M.M. Fleck, D.A. Forsyth, and C. Bregler. Finding Naked People. In Proc. Jth European Conf. on Computer Vision, pages 593-602, 1996.

....these, 20] who used elastic graphs to represent hands in different postures with local jets of Gabor filters computed at each vertex, 17] who detected maxima in a multi scale wavelet transform. The use of chromaticity as a primary cue for detecting skin coloured regions was first proposed by [5]. Our implementation of particle filtering largely follows the traditional approaches for condensation as presented by [8, 1, 18] and others. Using the hierarchical multi scale structure of the hand models, however, we extended the layered sampling approach from [19] 7 Summary We have ....

M. Fleck, D. Forsyth, and C. Bregler. Finding naked people. In Fourth European Conference on Computer Vision, pages II:593--602, Cambridge, UK, 1996.

....is isolated from the arms. This is to be expected because the grouping process assumes small changes from one element to its neighbor. This offers us a useful segmentation technique where the shirt front, left arm and right arm would emerge as separate groups. Work on human figure recognition [2] have shown the usefulness of such segmentation into parts. Figure 4 shows another type of grouping. Here, we have a simple repeating pattern. Traditionally, this would not be regarded as texture as it does not have significant 2D spatial extent. Instead, it is a distinctive element in the scene. ....

M. Fleck, D. Forsyth, and C. Bregler. "Finding Naked People". In To appear in ECCV96', 1996.

....still images is due to Jones and Rehg [6] In their case, a skin color model is learned from a huge collection of web images. A Bayesian classifier for skin color is then constructed which also incorporates a model of the non skin class. The approach relies on color alone. Fleck and Forsyth [4] and Wang et al. [13] propose systems for filtering adult images by finding naked people. In the approach by Fleck and Forsyth a combination of low level image filters is used combining skin color and texture features. In this paper, we introduce a generative skin patch model combining color and ....

Margaret M. Fleck, David A. Forsyth, and Chris Bregler. Finding naked people. In ECCV (2), pages 593--602, 1996.

....techniques in a video indexing context one should account for this limited applicability. In [23] people detection is taken one step further, detecting not only the head, but the whole human body, by detecting different components of the human body. A similar part based approach is followed in [13] to detect naked people. Obviously this method is suited for specific genres only. The auditory channel also provides strong clues for presence of people in video documents through speech in the segment. When layout segmentation has been performed, classification of the different signal segments ....

M. Fleck, D. Forsyth, and C. Bregler. Finding naked people. In European Conference on Computer Vision, volume 2, pages 593--602, Cambridge, UK, 1996.

No context found.

M. M. Fleck, D. A. Forsyth, and C. Bregler, Finding Naked People, 4th European Conference on Computer Vision, Springer, II:591-602 (1996).

No context found.

Fleck, M. M., Forsyth, D. A., and Bregler, C. Finding Naked People. ECCV, 1996, pages 593-602.

No context found.

M.M. Fleck, D.A. Forsyth, and C. Bregler, "Finding naked people," in European Conference on Computer Vision, Cambridge, UK, 1996, Vol. 2, pp. 593--602.

No context found.

Fleck, M. M., Forsyth, D. A., and Bregler, C. Finding Naked People. ECCV, 1996, pages 593-602.

No context found.

M. M. Fleck, D. A. Forsyth, and C. Bregler. Finding naked people. In Proc. 4'th European Confon Computer Vision, 1996.

No context found.

M. Fleck, D. Forsyth, and C. Bregler. Finding naked people. In ECCV, volume 2, pages 593--602, 1996.

No context found.

M. Fleck, D. Forsyth, and C. Bregler. Finding naked people. In Proc. European Conf. on Computer Vision, pages 593-- 602. B. Buxton, R. Cipolla, Springer-Verlag, Berlin, Germany, 1996.

No context found.

M. M. Fleck, D. A. Forsyth, and C. Bregler. Finding naked people. In Proc. 4'th European Confon Computer Vision, 1996.

No context found.

David A. Forsyth and Margaret Fleck, Finding Naked People, journal reviewing, 1996.

First 50 documents  Next 50

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