Skip to content

Yucheng-Ren/Cellari-Deeplearning-Course

 
 

Repository files navigation

Cellari-Deeplearning-Course

Our network is based on Segnet. The task is to segment malicious cell from medical images.

The dataset can be found here

reproduce the model

##1 stage

200 epochs image size (128, 128) learning rate 1e-4 batch size 50

accuracy on test data 0.77

2 stage

200 epochs image size (192, 192) learning rate 5e-5 batch size 5

accuracy on test data 0.84

3 stage

200 epochs image size (256, 256) learning rate 5e-5 batch size 5

accuracy on test data 0.86

4 stage

200 epochs image size (256, 256) learning rate 1e-5 batch size 5

accuracy on test data 0.87

About

Repository for the Cellari project in the Deep Learning Course at DTU

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages

  • Jupyter Notebook 98.8%
  • Python 1.2%