Skip to content

TF Implementation of Multiple Instance Learning FCN.

License

Notifications You must be signed in to change notification settings

adelabdelli/TensorFlow-MIL

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

13 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Deep Multiple Instance Learning

This repository contains a TensorFlow implementation of the CNN-MIL combination described in Classifying and Segmenting Microscopy Images with Deep Multiple Instance Learning from Brendan Frey`s lab.

Getting Started

Options

The following options are available for running the model:

  • -e, Number of epochs for which to train the model
  • -r, Specify the seed
  • -b, Batch size for training
  • -s, Where to save model
  • -m, Name of model
  • -t, Whether to train (1) or load model (0)

Datasets

The datasets.py file contains a small cluttered MNIST dataset. Each image is 72 x 72 pixels and contains four numbers: three are 0's, the other one is a number 1 - 9 excluding 0. The locations of these numbers in the image are semi-random. This dataset is a much smaller version of what the authors use in the paper.

About

TF Implementation of Multiple Instance Learning FCN.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 100.0%