Parallel-Hierarchical Neural Network for 3D Object Recognition

スポンサーリンク

概要

• 論文の詳細を見る

• In this paper, we propose a parallel-hierarchical neural network for 3D object recognition. This network is based on a visual nervous network having a hierarchical parallel structure and a recognition with the aid of memory. Structure of the network refers to Neocognitron which is based on a hierarchical parallel structure. The network transforms input image to match the learned image. The transformed image is repeatedly inputed to the network. This repetition process is based on a recognition with the aid of memory. By computer experiments using COIL-100, we confirm the effectiveness of the network for 3D object recognition. key words : 3D object recognition, affine transformation recognition, occlusion recognition, visual nervous network model

• 社団法人電子情報通信学会の論文
• 2003-04-01

著者

• Sato Noriaki

  Faculty of Science, Tohoku University

• Hagiwara Masafumi

  Faculty Of Science And Technology Keio University

• Sato Noriaki

  Faculty Of Science And Technology Keio University

関連論文

• High Field Magnetization of UPd_2Al_3
• Parallel-Hierarchical Neural Network for 3D Object Recognition
• A Multi-Winner Associative Memory
• Analysis of Momentum Term in Back-Propagation
• Quick Learning for Bidirectional Associative Memory (Special Issue on Neurocomputing)

スポンサーリンク