pierrehao / senet

Squeeze-and-Excitation Networks

Squeeze-and-Excitation Networks

By Jie Hu^[1], Li Shen^[2], Gang Sun^[1]. (arxiv)

Momenta^[1] and University of Oxford^[2].

Approach

Figure 1: Diagram of a Squeeze-and-Excitation building block.

 

Figure 2: Schema of SE-Inception and SE-ResNet modules.

Implementation

In this repository, Squeeze-and-Excitation Networks are implemented by Caffe.

Note:

• For efficient training and testing, we combine the consecutive operations channel-wise scale and element-wise summation into a single layer "Axpy" in the architectures with skip-connections, resulting in considerable memory and time comsuming reduce.

• Additonally, we found that the global average pooling implemented by cuDNN or BVLC/caffe is much slow on GPU. So we re-implement this operation on GPU and achieve a significant speedup.

Trained Models

Table 1. Single crop validation error on ImageNet-1k (center 224x224 crop from resized image with shorter side = 256). The SENet^* is one of superior models used in ILSVRC 2017 Image Classification Challenge where we won the 1st place (Team name: WMW).

Model	Top-1	Top-5	Size	Caffe Model
SE-BN-Inception	23.62	7.04	46 M	GoogleDrive
SE-ResNet-50	22.37	6.36	107 M	GoogleDrive
SE-ResNet-101	21.75	5.72	189 M	GoogleDrive
SE-ResNet-152	21.34	5.54	256 M	GoogleDrive
SE-ResNeXt-50 (32 x 4d)	20.97	5.54	105 M	GoogleDrive
SE-ResNeXt-101 (32 x 4d)	19.81	4.96	187 M	GoogleDrive
SENet*	18.68	4.47	440 M	GoogleDrive

Here we obtain better performances than those reported in the paper. We re-train all above models on a single GPU server equipped with 8 NVIDIA Titan X cards, using a mini-batch of 256 and a initial learning rate of 0.1 with more epoches. In our paper, we use large batch-size (1024) and learning rate (0.6).

Citation

If you use Squeeze-and-Excitation Networks in your research, please cite the paper:

@article{hu2017,
  title={Squeeze-and-Excitation Networks},
  author={Jie Hu and Li Shen and Gang Sun},
  journal={arXiv preprint arXiv:},
  year={2017}
}