Skip to content
residual-SqueezeNet
Branch: master
Clone or download
Song
Latest commit f8dc320 Mar 15, 2019
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
figure add fig Apr 22, 2016
stylesheets Create master branch via GitHub Oct 12, 2016
LICENSE Create LICENSE May 7, 2018
README.md Update README.md Mar 15, 2019
SqueezeNet_residual_top1_0.6038_top5_0.8250.caffemodel init Apr 22, 2016
index.html Create master branch via GitHub Oct 12, 2016
params.json Create master branch via GitHub Oct 12, 2016
trainval.prototxt init Apr 22, 2016

README.md

SqueezeNet-Residual

The repo contains the residual-SqueezeNet, which is obtained by adding bypass layer to SqueezeNet_v1.0. Residual-SqueezeNet improves the top-1 accuracy of SqueezeNet by 2.9% on ImageNet without changing the model size(only 4.8MB).

Related repo and paper

SqueezeNet

SqueezeNet-Deep-Compression

SqueezeNet-Generator

SqueezeNet-DSD-Training

SqueezeNet-Residual

If you find residual-SqueezeNet useful in your research, please consider citing the paper:

@article{SqueezeNet,
  title={SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and< 0.5MB model size},
  author={Iandola, Forrest N and Han, Song and Moskewicz, Matthew W and Ashraf, Khalid and Dally, William J and Keutzer, Kurt},
  journal={arXiv preprint arXiv:1602.07360},
  year={2016}
}

Usage

$CAFFE_ROOT/build/tools/caffe test --model=trainval.prototxt --weights=SqueezeNet_residual_top1_0.6038_top5_0.8250.caffemodel --iterations=1000 --gpu 0

Result

I0422 14:07:39.810755 32299 caffe.cpp:293] accuracy_top1 = 0.603759
I0422 14:07:39.810775 32299 caffe.cpp:293] accuracy_top5 = 0.824981
I0422 14:07:39.810792 32299 caffe.cpp:293] loss = 1.76711 (* 1 = 1.76711 loss) 

Architecture of the residual SqueezeNet


The building block:

You can’t perform that action at this time.