Semantic-segmentation-teeth-image

** Contributor
JongHwan Park : bomebug15@ds.seoultech.ac.kr (https://github.com/chromatices)

• fcn result image
  

• deeplab result image
  

• U net result image
  

Using only 50 teeth image and 50 mask image, we can check process of training normally

There's three main files. Choose what use model you want

main.py => Using U-net model

fcn_main.py => Using fcn model

deeplab_main.py => Using deeplab model

• each main files have 6 args parsers

mode

optim

loss-function

epochs

tricks

train/validation batches

example

$ python main.py --epoch 30 loss-function bce