CASIA OpenIR  > 智能感知与计算研究中心
Irregular Convolutional Neural Networks
Ma JB(马佳彬); Wang W(王威); Wang L(王亮)
Conference NameAsian Conference on Pattern Recognition (ACPR)
Conference DateNovember 26-29 2017
Conference PlaceNanjing, China
AbstractConvolutional kernels are basic and vital components of deep Convolutional Neural Networks (CNN). In this paper, we equip convolutional kernels with shape attributes to generate the deep Irregular Convolutional Neural Networks (ICNN). Compared to traditional CNN applying regular convolutional kernels like 3 × 3, our approach trains irregular kernel shapes to better fit the geometric variations of input features. In other words, shapes are learnable parameters in addition to weights. The kernel shapes and weights are learned simultaneously during end-to-end training with the standard back-propagation algorithm. Experiments for semantic segmentation are implemented to validate the effectiveness of our proposed ICNN
Indexed ByEI
Document Type会议论文
Recommended Citation
GB/T 7714
Ma JB,Wang W,Wang L. Irregular Convolutional Neural Networks[C],2017.
Files in This Item: Download All
File Name/Size DocType Version Access License
PID4986403.pdf(4971KB)会议论文 开放获取CC BY-NC-SAView Download
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[Ma JB(马佳彬)]'s Articles
[Wang W(王威)]'s Articles
[Wang L(王亮)]'s Articles
Baidu academic
Similar articles in Baidu academic
[Ma JB(马佳彬)]'s Articles
[Wang W(王威)]'s Articles
[Wang L(王亮)]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[Ma JB(马佳彬)]'s Articles
[Wang W(王威)]'s Articles
[Wang L(王亮)]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: PID4986403.pdf
Format: Adobe PDF
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.