Squeeze-and-Excitation Networks

, ,  -

Publications: arXiv Add/Edit

Abstract: Add/Edit

Convolutional neural networks are built upon the convolution operation, which extracts informative features by fusing spatial and channel-wise information together within local receptive fields. In order to boost the representational power of a network, much existing work has shown the benefits of enhancing spatial encoding. In this work, we focus on channels and propose a novel architectural unit, which we term the "Squeeze-and-Excitation" (SE) block, that adaptively recalibrates channel-wise feature responses by explicitly modelling interdependencies between channels. We demonstrate that by stacking these blocks together, we can construct SENet architectures that generalise extremely well across challenging datasets. Crucially, we find that SE blocks produce significant performance improvements for existing state-of-the-art deep architectures at slight computational cost. SENets formed the foundation of our ILSVRC 2017 classification submission which won first place and significantly reduced the top-5 error to 2.251%, achieving a 25% relative improvement over the winning entry of 2016.

Keywords: Add/Edit


Code Links

Languages: Cuda Add/Edit

Libraries: Add/Edit

Description: Add/Edit

Squeeze-and-Excitation Networks

0

Languages: Matlab Add/Edit

Libraries: Add/Edit

Description: Add/Edit

Memory consumption and FLOP count estimates for convnets

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

Implementation of SENets by chainer (Squeeze-and-Excitation Networks: https://arxiv.org/abs/1709.01507)

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

:fire::fire: A MXNet implementation of Squeeze-and-Excitation Networks (SE-ResNext, SE-Resnet, SE-Inception-v4 and SE-Inception-Resnet-v2) :fire::fire:

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

Squeeze-and-Excitation module for convolutional neural networks written in PyTorch

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

Naive implementation of SENet

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

SE-Net Incorporates with ResNet and WideResnet on CIFAR-10/100 Dataset.

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

List of books that I like related to deep learning

0

Languages: Jupyter Notebook Add/Edit

Libraries: Add/Edit

Description: Add/Edit

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

The solution for the Carvana Image Masking Challenge on Kaggle. It uses a custom version of RefineNet with Squeeze-and-Excitation modules implemented in PyTorch. It was a part of the final ensemble that was ranked 23 out of 735 teams (top 4%).

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

Implementation of IndRNN in Tensorflow

0

Languages: Matlab Add/Edit

Libraries: Add/Edit

Description: Add/Edit

Squeeze-and-Excitation Networks

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

Experimental keras implementation of novel neural network structures

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

Using cifar-10 datasets to learn deep learning.

0

Languages: Jupyter Notebook Add/Edit

Libraries: Add/Edit

Description: Add/Edit

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

Collection of Keras models used for classification

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

Implementation of Squeeze and Excitation Networks in Keras

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

A squeeze-and-excitation enabled ResNet for image classification

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

Deep Learning Architecture Genealogy Project

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

PyTorch implementation of SENet

0

Languages: Jupyter Notebook Add/Edit

Libraries: Add/Edit

Description: Add/Edit

Building MLP, Conv Net, LSTM, GAN, DCGAN models using low-level and high-level APIs

0

Languages: Python Add/Edit

Libraries: Add/Edit

Description: Add/Edit

implementation of squeze excitation moduels in keras

0