-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathrun_test_seg-unet.sh
25 lines (22 loc) · 1.11 KB
/
run_test_seg-unet.sh
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
export CUDA_VISIBLE_DEVICES=0
#i=0
k=10
img_size=768
model_name=unet-resnet18
norm_type=bn
saved_iter=best-dice
thresh=0.5
non_empty_ratio=1.0
root_ckpts=checkpoints_seg
root_results=results_seg
for((i=0;i<$k;i=i+1))
do
#python test.py --model_path ${root_ckpts}/${model_name}_siim_${saved_iter}_${img_size}x${img_size}_${i}-${k}_model.pth --norm_type ${norm_type} --dataset siim \
# --img_rows ${img_size} --img_cols ${img_size} --seed 1234 --fold_num $i --num_folds $k \
# --root_results ${root_results} \
# --thresh ${thresh} --non_empty_ratio ${non_empty_ratio} --batch_size 1 --split val --gt --only_non_empty #--use_cbam #--tta
python test.py --model_path ${root_ckpts}/${model_name}_siim_${saved_iter}_${img_size}x${img_size}_${i}-${k}_model.pth --norm_type ${norm_type} --dataset siim \
--img_rows ${img_size} --img_cols ${img_size} --seed 1234 --fold_num $i --num_folds $k \
--root_results ${root_results} \
--thresh ${thresh} --non_empty_ratio ${non_empty_ratio} --batch_size 1 --split test --only_non_empty #--use_cbam #--tta #--gt
done