- https://github.com/divamgupta/image-segmentation-keras 에서 개발한 pip package를 이용해서 image segmentation을 함 해 봄.
- training 은 하지 않고, 기본적으로 제공되는 3종의 pretrained 데이터를 이용해서 시험해 봄.
- ./images : input images
- ./output : segmented image will be located
음.. 학습 epoch가 모자랐던 것일까.. 데이터 개수의 한계? 아님 신경망 자체의 한계? 뭘까..


