This is a Caffe implementation of Squeeze-and-Excitation Networks (SENet). For details, please read the original slides:

• Squeeze-and-Excitation Networks

For offical implementations, please check this repo SENet.

Here we provide a pretrained SE-ResNet-50 model on ImageNet, which achieves slightly better accuracy rates than the original one reported in the official repo. You can use the official bvlc caffe to run this model without any modifications.

The top-1/5 accuracy rates by using single center crop (crop size: 224x224, image size: 256xN):

For your convenience, we also provide a link to this model on Baidu Disk.

• BGR mean values [103.94,116.78,123.68] are subtracted
• scale: 0.017 is used as std values for image preprocessing
• Images labels are the same as fb.resnet.torch. We also provide synset.txt, which can be found here.