ARM

  1. ARM Compute Library案例:

Raspberry Pi 3

  1. 用opencv在树莓派上部署SqueezeNet
  • 案例:用OpenCV3.3.0在Raspberry Pi 3上部署 pre-trained SqueezeNet Neural Network
  • 产品化程度:
    比较高,可以输出结果。
1
2
3
4
5
6
7
8
9
10
11
(1)在图像上绘出最高的预测分类标签和相应的概率
(2)将前五个结果和概率打印到终端
(3)将图像显示在屏幕上

[INFO] loading model...
[INFO] classification took 0.4432 seconds
[INFO] 1. label: drake, probability: 0.25705
[INFO] 2. label: goose, probability: 0.18581
[INFO] 3. label: black stork, probability: 0.10414
[INFO] 4. label: hornbill, probability: 0.074497
[INFO] 5. label: quail, probability: 0.051127
  1. 使用Raspberry Pi和预先训练的深度学习神经网络对输入图像进行分类
  1. 在Raspberry Pi 中运行Movidius Neural Compute Stick
  1. 使用Raspberry Pi 3进行目标检测

其他

  1. 通过SqueezeNet进行猫狗识别
  1. MXNet在Raspberry Pi上的实时对象检测
  1. 手机上的Squeezing Deep Learning
    链接:https://www.slideshare.net/anirudhkoul/squeezing-deep-learning-into-mobile-phones

Comments

⬆︎TOP