Need help with running the Predictions and Evaluation part of the RasterVision tutorial: TypeError: self._hds cannot be converted to a Python object for pickling

[cpuah]

Hi all.

Im currently running RasterVision on a windows laptop via VSCode

I already previously encountered the Windows numpy bug in the Training a Model Tutorial (#1950 (reply in thread)) and have already resolved it with the work around. Many thanks @AdeelH

I moved onto the next tutorial and am stuck with the following error on running this bit
pred_labels = SemanticSegmentationLabels.from_predictions(
ds.windows,
predictions,
smooth=True,
extent=ds.scene.extent,
num_classes=len(class_config))

**I understand that it is not optimal running RasterVision on a windows machine, but such are my circumstances at the moment.
I did not have much luck with docker. the image won't run telling me that The NVIDIA Driver was not detected. Installing via pip torch cuda 118 breaks what I have for rastervision.

The error message is attached below. Any help or feedback welcome. Appreciate your time.

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[21], [line 1](vscode-notebook-cell:?execution_count=21&line=1)
----> [1](vscode-notebook-cell:?execution_count=21&line=1) pred_labels = SemanticSegmentationLabels.from_predictions(
      [2](vscode-notebook-cell:?execution_count=21&line=2)     ds.windows,
      [3](vscode-notebook-cell:?execution_count=21&line=3)     predictions,
      [4](vscode-notebook-cell:?execution_count=21&line=4)     smooth=True,
      [5](vscode-notebook-cell:?execution_count=21&line=5)     extent=ds.scene.extent,
      [6](vscode-notebook-cell:?execution_count=21&line=6)     num_classes=len(class_config))

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\site-packages\rastervision\core\data\label\semantic_segmentation_labels.py:210](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:210), in SemanticSegmentationLabels.from_predictions(cls, windows, predictions, extent, num_classes, smooth, crop_sz)
    [184](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:184) """Instantiate from windows and their corresponding predictions.
    [185](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:185) 
    [186](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:186) Args:
   (...)
    [207](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:207)     Otherwise, a SemanticSegmentationDiscreteLabels.
    [208](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:208) """
    [209](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:209) labels = cls.make_empty(extent, num_classes, smooth=smooth)
--> [210](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:210) labels.add_predictions(windows, predictions, crop_sz=crop_sz)
    [211](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:211) return labels

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\site-packages\rastervision\core\data\label\semantic_segmentation_labels.py:237](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:237), in SemanticSegmentationLabels.add_predictions(self, windows, predictions, crop_sz)
    [233](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:233)     windows, predictions = discard_prediction_edges(
    [234](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:234)         windows, predictions, crop_sz)
    [235](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:235) # If predictions is tqdm-wrapped, it needs to be the first arg to zip()
    [236](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:236) # or the progress bar won't terminate with the correct count.
--> [237](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:237) for prediction, window in zip(predictions, windows):
    [238](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/core/data/label/semantic_segmentation_labels.py:238)     self[window] = prediction

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\site-packages\tqdm\notebook.py:254](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/tqdm/notebook.py:254), in tqdm_notebook.__iter__(self)
    [252](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/tqdm/notebook.py:252) try:
    [253](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/tqdm/notebook.py:253)     it = super(tqdm_notebook, self).__iter__()
--> [254](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/tqdm/notebook.py:254)     for obj in it:
    [255](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/tqdm/notebook.py:255)         # return super(tqdm...) will not catch exception
    [256](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/tqdm/notebook.py:256)         yield obj
    [257](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/tqdm/notebook.py:257) # NB: except ... [ as ...] breaks IPython async KeyboardInterrupt

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\site-packages\tqdm\std.py:1178](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/tqdm/std.py:1178), in tqdm.__iter__(self)
   [1175](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/tqdm/std.py:1175) time = self._time
   [1177](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/tqdm/std.py:1177) try:
-> [1178](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/tqdm/std.py:1178)     for obj in iterable:
   [1179](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/tqdm/std.py:1179)         yield obj
   [1180](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/tqdm/std.py:1180)         # Update and possibly print the progressbar.
   [1181](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/tqdm/std.py:1181)         # Note: does not call self.update(1) for speed optimisation.

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\site-packages\rastervision\pytorch_learner\learner.py:660](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:660), in <genexpr>(.0)
    [658](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:658) if numpy_out:
    [659](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:659)     if return_format == 'z':
--> [660](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:660)         preds = (self.output_to_numpy(p) for p in preds)
    [661](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:661)     else:
    [662](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:662)         # only convert z
    [663](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:663)         preds = ((*p[:-1], self.output_to_numpy(p[-1])) for p in preds)

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\site-packages\rastervision\pytorch_learner\learner.py:721](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:721), in <genexpr>(.0)
    [719](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:719)     preds = ((y, z) for _, y, z in preds)
    [720](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:720) elif return_format == 'z':
--> [721](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:721)     preds = (z for _, _, z in preds)
    [723](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:723) return preds

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\site-packages\rastervision\pytorch_learner\learner.py:758](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:758), in Learner._predict_dataloader(self, dl, raw_out, batched_output, predict_kw)
    [755](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:755) else:
    [756](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:756)     self.model.eval()
--> [758](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:758) for x, y in dl:
    [759](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:759)     if self.onnx_mode:
    [760](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:760)         z = self.predict_onnx(x, raw_out=raw_out, **predict_kw)

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\site-packages\torch\utils\data\dataloader.py:444](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:444), in DataLoader.__iter__(self)
    [442](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:442)     return self._iterator
    [443](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:443) else:
--> [444](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:444)     return self._get_iterator()

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\site-packages\torch\utils\data\dataloader.py:390](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:390), in DataLoader._get_iterator(self)
    [388](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:388) else:
    [389](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:389)     self.check_worker_number_rationality()
--> [390](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:390)     return _MultiProcessingDataLoaderIter(self)

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\site-packages\torch\utils\data\dataloader.py:1077](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:1077), in _MultiProcessingDataLoaderIter.__init__(self, loader)
   [1070](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:1070) w.daemon = True
   [1071](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:1071) # NB: Process.start() actually take some time as it needs to
   [1072](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:1072) #     start a process and pass the arguments over via a pipe.
   [1073](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:1073) #     Therefore, we only add a worker to self._workers list after
   [1074](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:1074) #     it started, so that we do not call .join() if program dies
   [1075](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:1075) #     before it starts, and __del__ tries to join but will get:
   [1076](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:1076) #     AssertionError: can only join a started process.
-> [1077](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:1077) w.start()
   [1078](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:1078) self._index_queues.append(index_queue)
   [1079](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:1079) self._workers.append(w)

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\multiprocessing\process.py:121](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/process.py:121), in BaseProcess.start(self)
    [118](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/process.py:118) assert not _current_process._config.get('daemon'), \
    [119](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/process.py:119)        'daemonic processes are not allowed to have children'
    [120](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/process.py:120) _cleanup()
--> [121](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/process.py:121) self._popen = self._Popen(self)
    [122](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/process.py:122) self._sentinel = self._popen.sentinel
    [123](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/process.py:123) # Avoid a refcycle if the target function holds an indirect
    [124](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/process.py:124) # reference to the process object (see bpo-30775)

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\multiprocessing\context.py:224](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/context.py:224), in Process._Popen(process_obj)
    [222](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/context.py:222) @staticmethod
    [223](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/context.py:223) def _Popen(process_obj):
--> [224](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/context.py:224)     return _default_context.get_context().Process._Popen(process_obj)

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\multiprocessing\context.py:327](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/context.py:327), in SpawnProcess._Popen(process_obj)
    [324](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/context.py:324) @staticmethod
    [325](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/context.py:325) def _Popen(process_obj):
    [326](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/context.py:326)     from .popen_spawn_win32 import Popen
--> [327](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/context.py:327)     return Popen(process_obj)

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\multiprocessing\popen_spawn_win32.py:93](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/popen_spawn_win32.py:93), in Popen.__init__(self, process_obj)
     [91](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/popen_spawn_win32.py:91) try:
     [92](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/popen_spawn_win32.py:92)     reduction.dump(prep_data, to_child)
---> [93](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/popen_spawn_win32.py:93)     reduction.dump(process_obj, to_child)
     [94](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/popen_spawn_win32.py:94) finally:
     [95](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/popen_spawn_win32.py:95)     set_spawning_popen(None)

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\multiprocessing\reduction.py:60](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/reduction.py:60), in dump(obj, file, protocol)
     [58](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/reduction.py:58) def dump(obj, file, protocol=None):
     [59](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/reduction.py:59)     '''Replacement for pickle.dump() using ForkingPickler.'''
---> [60](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/multiprocessing/reduction.py:60)     ForkingPickler(file, protocol).dump(obj)

File stringsource:2, in rasterio._io.DatasetReaderBase.__reduce_cython__()

TypeError: self._hds cannot be converted to a Python object for pickling

[AdeelH]

Try

from torch import multiprocessing

multiprocessing.set_start_method('fork')

If that doesn't help, disable multiprocessing in the data loader:

predictions = learner.predict_dataset(
    ...
    dataloader_kw=dict(num_workers=0),
)

[cpuah]

Thank you so much @AdeelH

The first bit didn't do much b/c RuntimeError: context has already been set

but the second bit got the script working to the end.

Many thanks!

C

[AdeelH]

Can you also try, instead of num_workers=0, this:

predictions = learner.predict_dataset(
    ...
    dataloader_kw=dict(multiprocessing_context='fork'),
)

?

[cpuah]

it didn't like it :)

ValueError: multiprocessing_context option should specify a valid start method in ['spawn'], but got multiprocessing_context='fork'

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[10], [line 1](vscode-notebook-cell:?execution_count=10&line=1)
----> [1](vscode-notebook-cell:?execution_count=10&line=1) predictions = learner.predict_dataset(
      [2](vscode-notebook-cell:?execution_count=10&line=2)     ds,
      [3](vscode-notebook-cell:?execution_count=10&line=3)     raw_out=True,
      [4](vscode-notebook-cell:?execution_count=10&line=4)     numpy_out=True,
      [5](vscode-notebook-cell:?execution_count=10&line=5)     predict_kw=dict(out_shape=(325, 325)),
      [6](vscode-notebook-cell:?execution_count=10&line=6)     #dataloader_kw=dict(num_workers=0), # line added per response for https://github.com/azavea/raster-vision/discussions/2064 
      [7](vscode-notebook-cell:?execution_count=10&line=7)     dataloader_kw=dict(multiprocessing_context='fork'),
      [8](vscode-notebook-cell:?execution_count=10&line=8)     progress_bar=True)

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\site-packages\rastervision\pytorch_learner\learner.py:649](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:649), in Learner.predict_dataset(self, dataset, return_format, raw_out, numpy_out, predict_kw, dataloader_kw, progress_bar, progress_bar_kw)
    [642](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:642) dl_kw = dict(
    [643](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:643)     collate_fn=self.get_collate_fn(),
    [644](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:644)     batch_size=self.cfg.solver.batch_sz,
    [645](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:645)     num_workers=self.cfg.data.num_workers,
    [646](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:646)     shuffle=False,
    [647](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:647)     pin_memory=True)
    [648](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:648) dl_kw.update(dataloader_kw)
--> [649](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:649) dl = DataLoader(dataset, **dl_kw)
    [651](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:651) preds = self.predict_dataloader(
    [652](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:652)     dl,
    [653](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:653)     return_format=return_format,
    [654](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:654)     raw_out=raw_out,
    [655](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:655)     batched_output=False,
    [656](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:656)     predict_kw=predict_kw)
    [658](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/rastervision/pytorch_learner/learner.py:658) if numpy_out:

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\site-packages\torch\utils\data\dataloader.py:262](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:262), in DataLoader.__init__(self, dataset, batch_size, shuffle, sampler, batch_sampler, num_workers, collate_fn, pin_memory, drop_last, timeout, worker_init_fn, multiprocessing_context, generator, prefetch_factor, persistent_workers, pin_memory_device)
    [260](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:260) self.timeout = timeout
    [261](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:261) self.worker_init_fn = worker_init_fn
--> [262](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:262) self.multiprocessing_context = multiprocessing_context
    [264](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:264) # Adds forward compatibilities so classic DataLoader can work with DataPipes:
    [265](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:265) #   _DataPipeSerializationWrapper container makes it easier to serialize without redefining pickler
    [266](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:266) if isinstance(self.dataset, IterDataPipe):

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\site-packages\torch\utils\data\dataloader.py:421](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:421), in DataLoader.__setattr__(self, attr, val)
    [417](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:417) if self.__initialized and attr in (
    [418](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:418)         'batch_size', 'batch_sampler', 'sampler', 'drop_last', 'dataset', 'persistent_workers'):
    [419](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:419)     raise ValueError(f'{attr} attribute should not be set after {self.__class__.__name__} is initialized')
--> [421](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:421) super().__setattr__(attr, val)

File [c:\Users\c_pua\AppData\Local\Programs\Python\Python310\lib\site-packages\torch\utils\data\dataloader.py:399](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:399), in DataLoader.multiprocessing_context(self, multiprocessing_context)
    [397](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:397)     valid_start_methods = multiprocessing.get_all_start_methods()
    [398](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:398)     if multiprocessing_context not in valid_start_methods:
--> [399](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:399)         raise ValueError(
    [400](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:400)             'multiprocessing_context option '
    [401](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:401)             f'should specify a valid start method in {valid_start_methods!r}, but got '
    [402](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:402)             f'multiprocessing_context={multiprocessing_context!r}')
    [403](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:403)     multiprocessing_context = multiprocessing.get_context(multiprocessing_context)
    [405](file:///C:/Users/c_pua/AppData/Local/Programs/Python/Python310/lib/site-packages/torch/utils/data/dataloader.py:405) if not isinstance(multiprocessing_context, python_multiprocessing.context.BaseContext):

ValueError: multiprocessing_context option should specify a valid start method in ['spawn'], but got multiprocessing_context='fork'

[AdeelH]

Ah, gotcha. Looks like fork isn't supported at all in Windows.