TCCNet

TCCNet: Temporally Consistent Context-Free Network for Semi-supervised Video Polyp Segmentation

1. datasets

• Train/Test datasets should be prepared in folder dataset
• Folder should be ordered as follows,

  
  |-- datasets
  |   |-- TrainSet
  |   |   |-- CVC-ClinicDB-612
  |   |   |   |-- 1,5,8,9,10,12,13,14,18,19,20,21,22,23,25,26,27,28
  |   |   |   |   |-- Frame
  |   |   |   |   |-- GT
  |   |   |   |   |-- border
  |   |   |-- CVC-colonDB-300
  |   |   |   |-- 0,1,2,4,9,10,11
  |   |   |   |   |-- Frame
  |   |   |   |   |-- GT
  |   |   |   |   |-- border

| |--TestSet | | |-- CVC-ClinicDB-612-Valid | | | |-- Frame | | | | |-- 2,3,4,11,15,17 | | | |-- GT | | | | |-- 2,3,4,11,15,17 | | |-- CVC-ClinicDB-612-Test | | | |-- Frame | | | | |-- 0,6,7,16,24 | | | |-- GT | | | | |-- 0,6,7,16,24 | | |-- CVC-colonDB-300 | | | |-- Frame | | | | |-- 3,5,6,7,8,12 | | | |-- GT | | | | |-- 3,5,6,7,8,12 | | |-- ETIS | | | |-- Frame | | | | |-- 0~25 | | | |-- GT | | | | |-- 0~25

## 2. Train & Test
- Run `python data/sqc_pathlist.py` to get `sqc_pathlist.npy`
- first stage `python pretraining.py`
- second stage `python main_training.py`
- test `python MyTest.py --load your_model`