Trouble in training on ssb-c . dataloader issue

[hyunjunChhoi]

Thanks for your great work

I had no problem running on DomainNet

However, I got some trouble on SSB-C benchmark

Can you help?

Running :

bash scripts/mi_pmtrans/ssbc.sh 0

Result:

cubc
Using evaluation function v2 to print results
/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torchvision/transforms/transforms.py:329: UserWarning: Argument 'interpolation' of type int is deprecated since 0.13 and will be removed in 0.15. Please use InterpolationMode enum.
warnings.warn(
Epoch: 0
0%| | 0/1778 [00:00<?, ?it/s]
Traceback (most recent call last):
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/runpy.py", line 197, in _run_module_as_main
return _run_code(code, main_globals, None,
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/runpy.py", line 87, in _run_code
exec(code, run_globals)
File "/home/hjchoi/HiLo/methods/ours/mi_dis_pm.py", line 363, in
train(model, dom_head, con_mlp, train_loader, optimizer, exp_lr_scheduler, cluster_criterion, mi_criterion, uda_criterion, mem_fea, mem_cls, epoch, args)
File "/home/hjchoi/HiLo/methods/ours/mi_dis_pm.py", line 134, in train
for batch_idx, batch in enumerate(tqdm(train_loader)):
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/tqdm/std.py", line 1182, in iter
for obj in iterable:
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/dataloader.py", line 628, in next
data = self._next_data()
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/dataloader.py", line 1333, in _next_data
return self._process_data(data)
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/dataloader.py", line 1359, in _process_data
data.reraise()
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/_utils.py", line 543, in reraise
raise exception
IndexError: Caught IndexError in DataLoader worker process 0.
Original Traceback (most recent call last):
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/_utils/worker.py", line 302, in _worker_loop
data = fetcher.fetch(index)
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/_utils/fetch.py", line 58, in fetch
data = [self.dataset[idx] for idx in possibly_batched_index]
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/_utils/fetch.py", line 58, in
data = [self.dataset[idx] for idx in possibly_batched_index]
File "/home/hjchoi/HiLo/data/data_utils.py", line 100, in getitem
img, label, uq_idx = self.unlabelled_dataset2[item - len(self.labelled_dataset) - len(self.unlabelled_dataset1)]
File "/home/hjchoi/HiLo/data/cubc.py", line 193, in getitem
sample = self.data.iloc[idx-((int(sev)-1)*len(self.data)*len(distortions)+(idx//len(self.data))*len(self.data))]
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/pandas/core/indexing.py", line 1103, in getitem
return self._getitem_axis(maybe_callable, axis=axis)
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/pandas/core/indexing.py", line 1656, in _getitem_axis
self._validate_integer(key, axis)
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/pandas/core/indexing.py", line 1589, in _validate_integer
raise IndexError("single positional indexer is out-of-bounds")
IndexError: single positional indexer is out-of-bounds

scarsc
Using evaluation function v2 to print results
/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torchvision/transforms/transforms.py:329: UserWarning: Argument 'interpolation' of type int is deprecated since 0.13 and will be removed in 0.15. Please use InterpolationMode enum.
warnings.warn(
Epoch: 0
0%| | 0/2466 [00:00<?, ?it/s]
Traceback (most recent call last):
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/runpy.py", line 197, in _run_module_as_main
return _run_code(code, main_globals, None,
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/runpy.py", line 87, in _run_code
exec(code, run_globals)
File "/home/hjchoi/HiLo/methods/ours/mi_dis_pm.py", line 363, in
train(model, dom_head, con_mlp, train_loader, optimizer, exp_lr_scheduler, cluster_criterion, mi_criterion, uda_criterion, mem_fea, mem_cls, epoch, args)
File "/home/hjchoi/HiLo/methods/ours/mi_dis_pm.py", line 134, in train
for batch_idx, batch in enumerate(tqdm(train_loader)):
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/tqdm/std.py", line 1182, in iter
for obj in iterable:
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/dataloader.py", line 628, in next
data = self._next_data()
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/dataloader.py", line 1333, in _next_data
return self._process_data(data)
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/dataloader.py", line 1359, in _process_data
data.reraise()
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/_utils.py", line 543, in reraise
raise exception
IndexError: Caught IndexError in DataLoader worker process 0.
Original Traceback (most recent call last):
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/_utils/worker.py", line 302, in _worker_loop
data = fetcher.fetch(index)
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/_utils/fetch.py", line 58, in fetch
data = [self.dataset[idx] for idx in possibly_batched_index]
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/_utils/fetch.py", line 58, in
data = [self.dataset[idx] for idx in possibly_batched_index]
File "/home/hjchoi/HiLo/data/data_utils.py", line 100, in getitem
img, label, uq_idx = self.unlabelled_dataset2[item - len(self.labelled_dataset) - len(self.unlabelled_dataset1)]
File "/home/hjchoi/HiLo/data/scarsc.py", line 162, in getitem
data_idx = self.data_idx[idx-((int(sev)-1)*len(self.data_idx)*len(distortions)+(idx//len(self.data_idx))*len(self.data_idx))]
IndexError: list index out of range

fgvcc
Using evaluation function v2 to print results
/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torchvision/transforms/transforms.py:329: UserWarning: Argument 'interpolation' of type int is deprecated since 0.13 and will be removed in 0.15. Please use InterpolationMode enum.
warnings.warn(
Epoch: 0
0%| | 0/1044 [00:00<?, ?it/s]
Traceback (most recent call last):
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/runpy.py", line 197, in _run_module_as_main
return _run_code(code, main_globals, None,
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/runpy.py", line 87, in _run_code
exec(code, run_globals)
File "/home/hjchoi/HiLo/methods/ours/mi_dis_pm.py", line 363, in
train(model, dom_head, con_mlp, train_loader, optimizer, exp_lr_scheduler, cluster_criterion, mi_criterion, uda_criterion, mem_fea, mem_cls, epoch, args)
File "/home/hjchoi/HiLo/methods/ours/mi_dis_pm.py", line 134, in train
for batch_idx, batch in enumerate(tqdm(train_loader)):
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/tqdm/std.py", line 1182, in iter
for obj in iterable:
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/dataloader.py", line 628, in next
data = self._next_data()
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/dataloader.py", line 1333, in _next_data
return self._process_data(data)
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/dataloader.py", line 1359, in _process_data
data.reraise()
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/_utils.py", line 543, in reraise
raise exception
IndexError: Caught IndexError in DataLoader worker process 0.
Original Traceback (most recent call last):
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/_utils/worker.py", line 302, in _worker_loop
data = fetcher.fetch(index)
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/_utils/fetch.py", line 58, in fetch
data = [self.dataset[idx] for idx in possibly_batched_index]
File "/home/hjchoi/anaconda3/envs/hilo/lib/python3.9/site-packages/torch/utils/data/_utils/fetch.py", line 58, in
data = [self.dataset[idx] for idx in possibly_batched_index]
File "/home/hjchoi/HiLo/data/data_utils.py", line 100, in getitem
img, label, uq_idx = self.unlabelled_dataset2[item - len(self.labelled_dataset) - len(self.unlabelled_dataset1)]
File "/home/hjchoi/HiLo/data/fgvcc.py", line 352, in getitem
ori_path, target = self.samples[index-((int(sev)-1)*len(self.samples)*len(distortions)+(index//len(self.samples))*len(self.samples))]
IndexError: list index out of range

[whj363636]

Hi @hyunjunChhoi

Sorry for late reply. I have simplified the data indexing for ssb-c datasets. You can have a try.

[hyunjunChhoi]

Hi @hyunjunChhoi

Sorry for late reply. I have simplified the data indexing for ssb-c datasets. You can have a try.

Thanks for your response

However ,

in fgvcc.py

    sev = str((index // (len(self.samples) * len(distortions))) + 1)
    dist = distortions[(index // len(self.samples)) % len(distortions)]
    sample_idx = index % len(self.samples)
    
     **Here we don't have ori path **->>>     _, file_name = os.path.split(ori_path)
    path = os.path.join(self.root, dist, sev, file_name)

--> make error
if I insert ori_path, target = self.samples[sample_idx] then is it okay??

in scars.py

now it is working well

But I have a question about implementation

SSB-C dataset has same number of samples as original CUB, car, Aircraft??

Only you insert different kind of distortion and noise severity ??

in cubc.py , and scars.py they refer to original root data sample number. Is it okay??

in cubc.py

    sev = str((idx // (len(self.data) * len(distortions))) + 1)
    dist = distortions[(idx // len(self.data)) % len(distortions)]
    data_idx = idx % len(self.data)
    
   **Here we don't have sample.filepath ** ->>   path = os.path.join(self.root, dist, sev, sample.filepath)
    target = sample.target - 1  # Targets start at 1 by default, so shift to 0
    img = self.loader(path)

    if self.transform is not None:
        img = self.transform(img)

    if self.target_transform is not None:
        target = self.target_transform(target)

---> make error

so if I insert sample = self.data.iloc[data_idx] in cubc.py
Is it working fine?? I am curious about that.
Because it has referred original root data sample number.

Can you check?

Thanks in advance.