CiteSeerX — Table search results

Results 1 - 10 of 39

Table 1 Taxonomy of neural networks for feature extraction (Mao and Jain, 1995)

by unknown authors

"... In PAGE 3: ... An arti5cial neural network model for conditional segmentation Neural networks as feature extraction and data pro- jection tools may be classi quot;ed according to both their mapping functions as linear and nonlinear ones as well as their learning methodology as supervised and unsuper- vised methods (Mao and Jain, 1995). For segmentation purposes, we typically consult unsupervised methods (see Table1 ); for discriminant analysis, on the other hand, supervised networks are applied. In the sequel, the com- ponents of our integrated approach will be shown and, quot;nally, the compositional methodology will be presented.... ..."

Table1: This table shows the Gibson mix ordered after the frequency of the instructions. This data was taken from [Jain91].

in Table of Contents

by Christoph Jechlitschek

TABLE V. LINES OF CODE (LOC) COMPARISON FOR DEVELOPING APPLICATIONS WITH/WITHOUT THE NCB ABSTRACTION. Application Based on JAIN

in A User-Centric Network Communication Broker for Multimedia Collaborative Computing

by Chi Zhang, S. Masoud Sadjadi, Weixiang Sun, Raju Rangaswami, Yi Deng

TABLE 3. Computer Time Requirements Amini, A., T. Weymouth, and R. Jain, 1990. Using dynamic program- ming for solving variational problems in vision, IEEE Trans. on Time (sec.) Synthetic Image Real Image

in Differential Snakes for Change Detection in Road Segments

by Peggy Agouris Anthony

Table 1: Besl and Jain apos;s segmenter parameters and values. Underlined values are the optimal values for our data set. An asterisk denotes the default value for each parameter. We trained with threshold values of 51%, 60%, 70%, 80% and 90%. The correct detections are out of 200 possible.

in Comparing Curved-Surface Range Image Segmenters

by Mark Powell Kevin, Kevin W. Bowyer Xiaoyi Jiang, Horst Bunke 1998

Cited by 13

Table 3: Recognition rates achieved using Jain and Zongker apos;s FS-SFFS algorithm with a k nearest neighbor classi er. No signi cant drop in performance occurs when E and R are eliminated from the initial pool.

in SmartCar: Detecting Driver Stress

by Jennifer Healey, Rosalind Picard 2000

Cited by 7

Table 1: Reading Schedule for Operating Systems course 66421 | OS refers to Applied Operating Systems Concepts by Silberschatz, Galvin and Gagne, PA refers to Jain apos;s The art of computer performance analysis and APU refers to Advanced Programming in the Unix Environment by Stevens

in unknown title

by unknown authors

"... In PAGE 4: ... 7 Schedule This schedule is tentative and is subject to change. The course schedule is described in Table1 . Project assignments and deadlines are to be announced during the semester.... ..."

Table 1: Besl and Jain apos;s segmenter parameters and values. Underlined values are the optimal values for our data set. An asterisk denotes the default value for each parameter. We trained with threshold values of 51%, 60%, 70%, 80% and 90%. There was a total of 40 regions in five images.

in Comparing Curved-Surface Range Image Segmentors

by Mark W. Powell, Mark Powell

"... In PAGE 6: ...LIST OF TABLES Table1... In PAGE 27: ... Each time the number of correct detections was increased by optimizing a parameter value, the amount of change is less than that of the previous parameter. Table1 is a list of all the parameters we trained and the values we chose for them. First, we varied the value of the RMS surface fit error threshold, which is sensitive to the degree of noise in the image and ranked highest in importance relative to performance.... ..."

Table 1: Results for two segmenters, one constructed from the region growing methods and one constructed from the unsupervised Bayesian classi cation methods, applied to the USF curved range data test set, along with the results of the University of Bern and Bessel and Jain range image segmentation routines. The results are presented for a comparison thresh- old of 51% (the minimum overlap of a machine segmented (MS) region with a ground-truth (GT) region for the former to be counted as part of the segmentation of the latter).

in Simple Surface Segmentation

by Dr Alan, Alan M. McIvor, David W. Penman, Peter T. Waltenberg 1997

"... In PAGE 4: ... This experiment mea- sures not only the ability of the algorithms to correctly segment regions of the given type but also their ability to reject regions of di erent type. Results are presented in Table1 . The poor performance in comparison to the other two methods is due to a number of reasons: 1.... ..."

Cited by 3

Table 1: Results for two segmenters, one constructed from the region growing methods and one constructed from the unsupervised Bayesian classi cation methods, applied to the USF curved range data test set, along with the results of the University of Bern and Bessel and Jain range image segmentation routines. The results are presented for a comparison thresh- old of 51% (the minimum overlap of a machine segmented (MS) region with a ground-truth (GT) region for the former to be counted as part of the segmentation of the latter).