The FaCells.

The FaCells. An Exploratory Study about LSTM Layers on Face Sketches Classifiers. Code implementing the paper https://arxiv.org/abs/2102.11361.

Try the demo in http://www.columbia.edu/~xig2000/FaCells/

To install the repo:

1. uncompress the data folder data file in the root directory, total 15.1 Gb. It will create a folder named 'data' with the input draws, the model, and previously calculated predictions.
2. install requirements.txt
3. see the initial_script.ipynb with the following lines.

from functions.paperFunctions import *
from functions.modelExploring import *
from functions.FaCells import FaCellsProject

facellsProject = FaCellsProject('absolute-sorted','data')
facellsProject.loadModel("modelX.h5", 'data')
facellsProject.loadPredictions()
#facellsProject.model.summary()

#plotOneFeature (featureId, numberOfDrawsToOverlap, threshold, facellsProject)
plotOneFeature(6,1000, 8, facellsProject)

the script will create the photo '6-1000-8.pdf' in the output dir