raise ValueError(f"Sequence must have length {dim}, got {len(tup)}.")

[mahootiha-maryam]

I use unet with this metrics as my model:

model = UNet(
    spatial_dims=3,
    in_channels=2,
    out_channels=2,
    channels=(16, 32, 64, 128, 256),
    strides=(2, 2, 2, 2),
    num_res_units=2,
    norm=Norm.BATCH,
).to(device)

I have train dataset That has 2 input channel image and 2 output channel label
And also I have test dataset That has 2 input channel image and 2 output channel label
These are the transformers for test :

test_transforms = Compose(
    [
        LoadImaged(keys=['image']),
        #add channels to the images that monai can interpret in this way
        AddChanneld(keys=['image']),
        # EnsureChannelFirstd(keys="image"),
        #rescale the voxels pixdim=(height,width,depth) 
        Spacingd(keys=['image'], pixdim=(1.5, 1.5, 2)),
        Orientationd(keys=['image'], axcodes="RAS"),
        ScaleIntensityRanged(keys=['image'], a_min=-100, a_max=1000, b_min=0.0, b_max=1.0, clip=True),
        Resized(keys=['image'], spatial_size=[128, 128, 128]),
        #to tensor should be the last transform
        ToTensord(keys=['image'])
    ]
)

And these are the post transformers:

post_transforms = Compose([
    Invertd(
        keys="pred",
        transform=test_transforms,
        orig_keys="image",
        meta_keys="pred_meta_dict",
        orig_meta_keys="image_meta_dict",
        meta_key_postfix="meta_dict",
        nearest_interp=False,
        to_tensor=True,
    ),
    AsDiscreted(keys="pred", argmax=True, to_onehot=None),
    SaveImaged(keys="pred", meta_keys="pred_meta_dict", output_dir="./out", output_postfix="seg", resample=False),
])

When I try to load a pretrained model and use test dataset for seeing the prediction with this code:

model.load_state_dict(torch.load(os.path.join(root_dir, "best_metric_model.pth")))
model.eval()

with torch.no_grad():
    for test_data in test_loader:
        print(test_data.shape)
        test_inputs = test_data[0].to(device)
        roi_size = (160, 160, 160)
        sw_batch_size = 4
        # test_data["pred"] 
        x = sliding_window_inference(test_inputs, roi_size, sw_batch_size, model)
        test_data = [post_transforms(i) for i in decollate_batch(test_data)]

I face this error on the line x = sliding... and when I tried to see the source of error I couldn't understand it.
```
raise ValueError(f"Sequence must have length {dim}, got {len(tup)}.")

ValueError: Sequence must have length 2, got 3.

Do you think the reason is that my images are nifti and we can not save images as nifti? If this is not reason can you please help me

[KumoLiu]

Hi @mahootiha-maryam,
It seems you lost your batch dimension when using sliding_window_inference. The code you post below took the first data of the batch. Maybe you can check the dimension of your test_inputs, its shape should be 'NCHW[D]'. Thanks.

test_inputs = test_data[0].to(device)