Training in 2D images

[noseDewdrop]

Thanks for ur great job, I'm trying apply it on 2D images which were created by 3D datas of Brats2019. All of them has the shape of (160, 160). Therefore, I set them in the same folder.Dataset is ok, but I got this error when I run the main.py.

ValueError: input image size (img_size) should be divisible by stage-wise image resolution.

Here's the output on command .

python3.8 main.py --feature_size=48 --batch_size=1 --logdir=unetr_test_dir --fold=1 --optim_lr=1e-4 --lrschedule=warmup_cosine --infer_overlap=0.5 --save_checkpoint --val_every=10 --json_list='./jsons/brats21_folds.json' --data_dir=/mnt/sdb_newdisk/ljc_cnu/circle_cut/Tasks/MICCAI/ --use_checkpoint --noamp

0 gpu 0 Batch size is: 1 epochs 300 Traceback (most recent call last): File "main.py", line 230, in main() File "main.py", line 104, in main main_worker(gpu=0, args=args) File "main.py", line 130, in main_worker model = SwinUNETR( File "/home/ljc_cnu/anaconda3/envs/swin/lib/python3.8/site-packages/monai/networks/nets/swin_unetr.py", line 111, in init raise ValueError("input image size (img_size) should be divisible by stage-wise image resolution.") ValueError: input image size (img_size) should be divisible by stage-wise image resolution.

can u give me a solution?

[noseDewdrop]

That's the result after I change the roi from 96 to 48, maybe that's wrong. But I got the another error when I set that with 96.

Traceback (most recent call last): File "/home/ljc_cnu/anaconda3/envs/swin/lib/python3.8/site-packages/torch/utils/data/_utils/worker.py", line 302, in _worker_loop data = fetcher.fetch(index) File "/home/ljc_cnu/anaconda3/envs/swin/lib/python3.8/site-packages/torch/utils/data/_utils/fetch.py", line 49, in fetch data = [self.dataset[idx] for idx in possibly_batched_index] File "/home/ljc_cnu/anaconda3/envs/swin/lib/python3.8/site-packages/torch/utils/data/_utils/fetch.py", line 49, in data = [self.dataset[idx] for idx in possibly_batched_index] File "/home/ljc_cnu/anaconda3/envs/swin/lib/python3.8/site-packages/monai/data/dataset.py", line 105, in getitem return self._transform(index) File "/home/ljc_cnu/anaconda3/envs/swin/lib/python3.8/site-packages/monai/data/dataset.py", line 91, in _transform return apply_transform(self.transform, data_i) if self.transform is not None else data_i File "/home/ljc_cnu/anaconda3/envs/swin/lib/python3.8/site-packages/monai/transforms/transform.py", line 118, in apply_transform raise RuntimeError(f"applying transform {transform}") from e RuntimeError: applying transform <monai.transforms.compose.Compose object at 0x7fc9c22c4940>

[marvnmtz]

Thanks for ur great job, I'm trying apply it on 2D images which were created by 3D datas of Brats2019. All of them has the shape of (160, 160). Therefore, I set them in the same folder.Dataset is ok, but I got this error when I run the main.py.

ValueError: input image size (img_size) should be divisible by stage-wise image resolution.

Here's the output on command .

python3.8 main.py --feature_size=48 --batch_size=1 --logdir=unetr_test_dir --fold=1 --optim_lr=1e-4 --lrschedule=warmup_cosine --infer_overlap=0.5 --save_checkpoint --val_every=10 --json_list='./jsons/brats21_folds.json' --data_dir=/mnt/sdb_newdisk/ljc_cnu/circle_cut/Tasks/MICCAI/ --use_checkpoint --noamp

0 gpu 0 Batch size is: 1 epochs 300 Traceback (most recent call last): File "main.py", line 230, in main() File "main.py", line 104, in main main_worker(gpu=0, args=args) File "main.py", line 130, in main_worker model = SwinUNETR( File "/home/ljc_cnu/anaconda3/envs/swin/lib/python3.8/site-packages/monai/networks/nets/swin_unetr.py", line 111, in init raise ValueError("input image size (img_size) should be divisible by stage-wise image resolution.") ValueError: input image size (img_size) should be divisible by stage-wise image resolution.

can u give me a solution?

I don't know if you still have that issue, but I would try to crop the images to a multiple of 96, so for example 192 as it is the closest to 160.

[Saillxl]

@noseDewdrop hello，do you remember how to solve this problem?