Skip to content

MakoOfficial/MyUnet

Repository files navigation

Run

python pretrain.py \
        --use_canny 0 \
        --data_path ../maske/train/ \
        --save_pth  ./checkpoint/checkpoint_ori_200.pth \
        --batch_size 20 \
        --epochs 200 \
        --save_ckpt_freq 200 \
        --input_size 224 \
        --lr 1e-4 \
        --step_size 50 \
        --gamma 0.7

local run ori

python pretrain.py
        --use_canny 0
        --data_path ../archive/masked_1K_train/ori
        --save_path  ./checkpoint/masked_1K_ori_200.pth
        --batch_size 8
        --epochs 200
        --save_ckpt_freq 200
        --input_size 512
        --lr 1e-4
        --step_size 50
        --gamma 0.7

local run canny

python pretrain.py
        --use_canny 1
        --data_path ../archive/masked_1K_train/canny
        --save_path  ./checkpoint/masked_1K_canny_200.pth
        --batch_size 8
        --epochs 200
        --save_ckpt_freq 200
        --input_size 512
        --lr 1e-4
        --step_size 50
        --gamma 0.7

server run ori

python pretrain.py \
        --use_canny 0 \
        --data_path ../masked_1K_train/ori \
        --save_path  ./checkpoint/masked_1K_ori_200.pth \
        --batch_size 8 \
        --epochs 200 \
        --save_ckpt_freq 200 \
        --input_size 512 \
        --lr 1e-4 \
        --step_size 50 \
        --gamma 0.7

server run canny

python pretrain.py \
        --use_canny 1 \
        --data_path ../masked_1K_train/canny \
        --save_path  ./checkpoint/masked_1K_canny_200.pth \
        --batch_size 8 \
        --epochs 200 \
        --save_ckpt_freq 200 \
        --input_size 512 \
        --lr 1e-4 \
        --step_size 50 \
        --gamma 0.7

if input_size = 512, on 12G memory, the batch is no more than 8

run demo ori

python demo.py
    --chkpt ./checkpoint/masked_1K/masked_1K_ori_200.pth
    --data_path ../archive/masked_crop/val/a
    --save_path ./output/masked_1K/masked_ori.png
    --input_size 512

run demo canny

python demo.py
    --chkpt checkpoint/masked_1K/masked_1K_canny_200.pth
    --data_path ../archive/masked_crop/canny/val
    --save_path ./output/masked_1K/masked_canny.png
    --input_size 512

ori2canny

python demo.py
    --chkpt ./checkpoint/masked_1K/masked_1K_ori_200.pth
    --data_path ../archive/masked_crop/canny/val
    --save_path ./output/masked_1K/ori2canny.png
    --input_size 512

canny2ori

python demo.py
    --chkpt checkpoint/shortcut/checkpoint_canny_200.pth
    --data_path ../archive/masked_crop/val/a
    --save_path ./output/shortcut/canny2ori.png
    --input_size 512

run class.py

python class.py \
    --batch_size 80 \
    --epochs 100 \
    --save_ckpt_freq 50 \
    --ori_ckpt_path ./checkpoint/masked_1K/masked_1K_ori_200.pth \
    --canny_ckpt_path ./checkpoint/masked_1K/masked_1K_canny_200.pth \
    --lr 1e-3 \
    --weight_decay 1e-5 \
    --step_size 30 \
    --gamma  0.5

run class2.py

python class2.py \
    --batch_size 20 \
    --epochs 100 \
    --save_ckpt_freq 50 \
    --ori_ckpt_path ./checkpoint/masked_1K/masked_1K_ori_200.pth \
    --canny_ckpt_path ./checkpoint/masked_1K/masked_1K_canny_200.pth \
    --lr 1e-3 \
    --weight_decay 1e-2 \
    --step_size 10 \
    --gamma  0.7

run K-fold.py

python K-fold.py \
    --batch_size 32 \
    --epochs 100 \
    --save_ckpt_freq 50 \
    --ori_ckpt_path ./checkpoint/masked_1K_ori_200.pth \
    --canny_ckpt_path ./checkpoint/masked_1K_canny_200.pth \
    --lr 1e-3 \
    --weight_decay 1e-4 \
    --step_size 20 \
    --gamma  0.6 \
    --ori_train_path ../masked_4K/masked/ori \
    --canny_train_path ../masked_4K/masked/canny 

run K-fold-MMANet.py

python K-fold-MMANet.py \
    --batch_size 32 \
    --epochs 50 \
    --save_ckpt_freq 50 \
    --lr 5e-4 \
    --weight_decay 0 \
    --step_size 10 \
    --gamma  0.5 \
    --ori_train_path ../../autodl-tmp/ori \
    --seed 1

python onlyUseOri.py
--batch_size 80
--epochs 200
--save_ckpt_freq 50
--ori_ckpt_path ./checkpoint/masked_1K/masked_1K_ori_200.pth
--canny_ckpt_path ./checkpoint/masked_1K/masked_1K_canny_200.pth
--lr 1e-3
--weight_decay 1e-5
--step_size 30
--gamma 0.5

python onlyUseOri.py 5e-4 128 100 1

python onlyUseCanny.py
--batch_size 80
--epochs 100
--save_ckpt_freq 50
--ori_ckpt_path ./checkpoint/masked_1K/masked_1K_ori_200.pth
--canny_ckpt_path ./checkpoint/masked_1K/masked_1K_canny_200.pth
--lr 1e-3
--weight_decay 1e-5
--step_size 30
--gamma 0.5

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages