Presentation 2012-12-12
Estimate the gradients of skew symmetric shapes using asymmetry of the shapes on a neural network model
Takashi Sugiura, Kazushi Murakoshi,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) It is possible for humans to estimate the gradients of the depth direction of the object in monocular. Some researchers proposed methods to estimate gradients. However, these methods were just based on image processing. In this paper, we propose a method to estimate the gradients close to human visual processing. For verification of method, we estimated the gradients of skew symmetric shapes (symmetric shape tilted in the direction of depth). This model extract the symmetry axis by neural network model. It regards the degree of asymmetry of the outline of shape as features. It calculated the features on square which rotated around the horizontal axis and vertical axis. Then, it learned the features by multilayer perceptron using error back propagation method. Thereafter, it estimated the gradients of rectangle after learning square. As a result, this model could estimate the gradients at $8.75$ degrees average error.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) estimate the gradients / skew symmetric shapes / asymmetry of the outline
Paper # NC2012-84
Date of Issue

Conference Information
Committee NC
Conference Date 2012/12/5(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Estimate the gradients of skew symmetric shapes using asymmetry of the shapes on a neural network model
Sub Title (in English)
Keyword(1) estimate the gradients
Keyword(2) skew symmetric shapes
Keyword(3) asymmetry of the outline
1st Author's Name Takashi Sugiura
1st Author's Affiliation Toyohashi University of Technology()
2nd Author's Name Kazushi Murakoshi
2nd Author's Affiliation Toyohashi University of Technology
Date 2012-12-12
Paper # NC2012-84
Volume (vol) vol.112
Number (no) 345
Page pp.pp.-
#Pages 4
Date of Issue