Jeff: fast.ai lesson 1&2感觉没讲什么干货。

~/keras/keras.json配置后台theano or tensorflow.

~/./theanorc处理器配置cpu or gpu.

[Model] VGG16-LMLPHP[Model] VGG16-LMLPHP

VGG16的预测

VGG16的训练

VGG16结构分析

  • GoogleNet略胜一筹

[Model] VGG16-LMLPHP

  • 结构还算简单。

[Model] VGG16-LMLPHP

小卷积核3*3:

  • Stack of three 3x3 conv (stride 1) layers has same effective receptive field as one 7x7 conv layer. 【第二层等价于5*5,第三层等价于7*7】
  • But deeper, more non-linearities.
  • And fewer parameters: 3 * (3C) vs.7C for C channels per layer.

[Model] VGG16-LMLPHP

VGG性价比不怎么高感觉。

[Model] VGG16-LMLPHP

感觉googLeNet更适合移动设备,结构有点小复杂,但高效。

[Model] VGG16-LMLPHP

Tensorflow VGG16 and VGG19

Ref: https://github.com/machrisaa/tensorflow-vgg

Ref: models/research/slim/nets/vgg.py ---- 需要学会使用官网代码!

05-11 11:25