Skip to content

Project of DataSci track of HackRice 17 (but forgot to submit)

Notifications You must be signed in to change notification settings

xun6000/hackrice17-KobeTeam

 
 

Repository files navigation

Identify the Painting with Deep Learning

  1. The general approach

In order to give a probability whether the two paintings are painted by the same artist, from the view of machine learning, we need a model which the inputs are the pair of paintings and the output ideally is a number indicates a probability (target output).

  1. Pre-trained model Using Inception v3 as a pretrained model, subsitute the original output layer with a single output sigmoid layer yields the probability.

  2. Randomly sample negative and positive examples, with 1 : 1 or 2 : 1 ratio, labels them with 1 or 0. Then train it.

About

Project of DataSci track of HackRice 17 (but forgot to submit)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 75.0%
  • Python 25.0%