【深度学习-模型系列】Inception

简介

• [v1 - 2014] Going Deeper with Convolutions
• [v2 - 2015] Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift 加了BatchNormalization正则，去除55卷积，用两个33代替
• [v3 - 2015] Rethinking the Inception Architecture for Computer Vision 77卷积又拆成71+1*7
• [v4 - 2016] Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning 加入了残差结构 code-tensorflow
  • 尽管Inception v3与Inception-ResNet-v2都很擅长识别每一条狗所属的类别，这种新模型的性能更为显著。例如，旧模型可能会错误地将右边的图片识别为阿拉斯加雪橇犬，新模型Inception-ResNet-v2能够准确地识别两个图片中狗的类别。

创新点

code

https://github.com/tensorflow/models/tree/master/research/inception