UNet from scratch (pytorch)
(gpu 장비가 없어서) training 중간에 멈춘 결과입니다.
├── UNet
├── datasets
│ └── cars
├── data
│ ├── __init__.py
│ └── dataset.py
├── option.py
├── model.py
├── train.py
├── inference.py
└── README.mddata/__init__.py: dataset loaderdata/dataset.py: data preprocess & get itemmodel.py: Define block and construct Modeloption.py: Environment setting
# Clone this repo and install dependency
git clone https://github.com/inhopp/UNet.gitpython3 train.py
--device {}(defautl: cpu) \
--data_name {}(default: cars) \
--lr {}(default: 0.0001) \
--n_epoch {}(default: 5) \
--num_workers {}(default: 2) \
--batch_size {}(default: 16) \
--eval_batch_size {}(default: 16)python3 inference.py
--device {}(defautl: cpu) \
--data_name {}(default: cars) \
--num_workers {}(default: 2) \
--eval_batch_size {}(default: 16)
https://github.com/aladdinpersson/Machine-Learning-Collection