Skip to content

Latest commit

 

History

History
55 lines (37 loc) · 930 Bytes

README.md

File metadata and controls

55 lines (37 loc) · 930 Bytes

Mahjong-Dection

This project is about Mahjong classification using keras.

Author:

  • Holman Tai
  • Johnman Li

Dependency

  1. selenium
  2. keras
  3. python3

This project contain two main parts.

  1. Data collection
  2. Vgg16 transfer learning

Part 1 - Data collection

searchPage_spide.py

search key word in taobao

itemSpider.py

save target img in description

scraperImg.py

download the img from the url saved

dataSetCropper.py

Crop the data base on the labbeling in xml file For your own data, please labbel it by your own


Part 2 - Keras transfer

Mahjong Classification.ipynb

ipython notebook to train a vgg16 network to this task

Test Model.ipynb

ipython notebook to load the model and run test imgs.


Sample images

Alt text Alt text Alt text Alt text Alt text