nutszebra/googlenet_v2

Implementation of googlenet-v2 by chainer (Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift: https://arxiv.org/abs/1502.03167)

What's this

Implementation of GoogLeNet-v2 [1] by chainer

Dependencies

git clone https://github.com/nutszebra/googlenet_v2.git
cd googlenet_v2
git submodule init
git submodule update

How to run

python main.py -p ./ -g 0 

Details about my implementation

• Data augmentation
  Train: Pictures are randomly resized in the range of [256, 512], then 224x224 patches are extracted randomly and are normalized locally. Horizontal flipping is applied with 0.5 probability.
  Test: Pictures are resized to 384x384, then they are normalized locally. Single image test is used to calculate total accuracy.

• Auxiliary classifiers
  No implementation

• Learning rate
  The description about the schedule of learning rate can't be found in [1], so as [2] said, learning rate are multiplied by 0.96 at every 8 epochs. Initial learning rate is 0.045 acoording to [1].

• Weight decay
  The description about weight decay can't be found in [1], so weight decay is 4.0*10^-5 as [3] says.

• Separable conv
  Normal convolution is used.

Cifar10 result

network	depth	total accuracy (%)
my implementation	32	94.89

References

Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift [1]
Going Deeper with Convolutions [2]
Rethinking the Inception Architecture for Computer Vision [3]