Skip to content

the python source code of JDD (Jing Dong Discovery) pig recognition contest.

Notifications You must be signed in to change notification settings

hy-2013/jdd_pig_recognition_contest

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

jdd_pig_recognition_contest

The python source code of JDD (Jing Dong Discovery) pig recognition contest (https://jddjr.jd.com/item/4).

1. 软件环境

  • tensorflow:0.10.0rc0
  • keras:2.1.2
  • opencv:3.3.1
  • pandas:0.19.2

2. 执行流程

  • video2image.py:实现猪mp4 videos到images的转化,以及将images分为train set和valid set。
  • jdd_pig_recognition.py:首先,通过vgg16生成train set、valid set和test set的1000维特征;然后构造dnn做多分类(30类);最后输出test set的预测结果。

3. 经验总结

  • 重要参数:样本量(包括train和valid样本的比例,若样本量够大,10:1或train更多)、epoch+batch(epoch一定要做到使训练指标稳定下来,batch的可尝试4、16、64、128、256等)、learning rate、optimizer等。

About

the python source code of JDD (Jing Dong Discovery) pig recognition contest.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages