Wrinkle_force_microscopy

For the review of Article "Wrinkle force microscopy: a new machine learningbased approach to predict cell mechanics from images".

A Graphic card with at least 6GB VRAM is recommended.

Environment:

  with Python 3.5 and CUDA 9.0:
        Tensorflow 1.13.0rc2,
        numpy 1.16.4,
        matplotlib 3.0.2,
        opencv-python 4.1.0.25

unzip data.zip will acquire training data.

Please unzip the items to 'data/' not 'data/data'
The data is divided into 5 parts, set the trigger '--dataset' in train.py and test.py to train or test it.
In a set of training folder, the train folder is for training, the val folder is for testing. 
Test data in the folder 'data/w2f_1/val' is well prepaired, if you want to test others, please prepair the data refer to it. 

The value of the force and the visualization of it can be acquired at the same time by running test.py.