/Squeeze-Excitation-PyTorch

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

Primary LanguagePython

Squeeze-and-Excitation

A Squeeze-and-Excitation module for convolutional neural networks written in PyTorch. Based on the paper "Squeeze-and-Excitation Networks" by Jie Hu, Li Shen and Gang Sun using global information for feature recalibration. The networks.py file contains the SqEx module, which performs the Squeeze-and-Excitation. The ResBlockSqEx module is an example of how to use the SqEx module.