About

This is my implementation of GoogLeNet Inception V3 by Pytorch.

The implementation can be adapted to any size of picture.

I just removed the dropout layer and changed it to a reeling down-dimensional layer. Others is almost same as the original network.

I think my implementation is clearer than the one in torchvision.

I've added some comments to my code that can help you better understand the network, as well as my code.

If my code can help you, then I'm really honored.

Dependency

I didn't test it in other environments, but in my following environment the network can work.

Python 3.6.9

​ -- Pytorch 1.3.1

Cuda 10.1.105

​ -- CuDNN 7.6.5

Ubuntu 16.04.7 LTS

Usage

Place the code in your project, and just import to it.

from googlenet_v3 import GoogLeNetV3
net = GoogLeNetV3(channels_in)

License

APACHE LICENSE, VERSION 2.0