Grouped Pointwise Convolutions Reduce Parameters in Convolutional Neural Networks

In DCNNs, the number of parameters in pointwise convolutions rapidly grows due to the multiplication of the number of filters by the number of input channels that come from the previous layer. Our proposal makes pointwise convolutions parameter efficient via grouping filters into parallel branches...

Full description

Bibliographic Details
Published in:Mendel
Main Authors: Joao Paulo Schwarz Schuler, Santiago Romani, Mohamed Abdel-Nasser, Hatem Rashwan, Domenec Puig
Format: Article
Language:English
Published: Brno University of Technology 2022-06-01
Subjects:
EfficientNet
Deep Learning
Computer Vision
CNN
DCNN
Online Access:https://mendel-journal.org/index.php/mendel/article/view/169

Internet

https://mendel-journal.org/index.php/mendel/article/view/169

Similar Items