论文An Analysis of Deep Neural Network Models for Practical Applications 比较了2016年前的一些神经网络的大小，精度等参数。如下图所示：
2016年以来有出现了一些新的神经网络结构，特地查阅了一些论文，总结了一下（发现有很多数据没法不全面或者不同版本的数据稍有差异，其中必定有很多纰漏之处，望多多指正，我会修改并补充的），如下：
其中包含～的数据是从An Analysis of Deep Neural Network Models for Practical Applications 文中图片得到的大概值。
参考文献
1. An Analysis of Deep Neural Network Models for Practical Applications
2. ImageNet Classification with Deep Convolutional Neural Networks
2. Very Deep Convolutional Networks for Large-Scale Image Recognition
3. Network In Network
4. Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification
5. Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
6. Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning
7. Deep Residual Learning for Image Recognition
8. Identity Mappings in Deep Residual Networks
8. Highway Networks
9. FractalNet: Ultra-Deep Neural Networks without Residuals
10. Deep Networks with Stochastic Depth
11. Aggregated Residual Transformations for Deep Neural Networks
12. Wide Residual Networks
13. Densely Connected Convolutional Networks(CVPR2017 best paper)
14. Deep Pyramidal Residual Networks
15. Dual Path Networks
16. SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size
17. MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
18. ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices