Add cortical layers dataset + support for 1D datasets#5
Conversation
fvisin
changed the title
|
Thank you for the PR! I will be traveling in the next 2 weeks, but I will try my best to look into it in a reasonable amount of time! |
c6a8d70 to
ff0bbfe
Compare
fvisin
left a comment
fvisin
left a comment
There was a problem hiding this comment.
Sorry for the long delay, I finally found time to review the PR! :)
| from matplotlib import cm | ||
|
|
||
| from dataset_loaders.parallel_loader import ThreadedDataset | ||
| # from parallel_loader_1D import ThreadedDataset_1D |
There was a problem hiding this comment.
Please remove the commented line.
| floatX = 'float32' | ||
|
|
||
| class CorticalLayersDataset(ThreadedDataset): | ||
| '''The Cortical Layers dataset consists of a number of 1D intensity profiles |
There was a problem hiding this comment.
Please comply with PEP257: add a short description in the first line followed by a blank line.
PEP257:
Multi-line docstrings consist of a summary line just like a one-line docstring, followed by a blank line, followed by a more elaborate description. The summary line may be used by automatic indexing tools; it is important that it fits on one line and is separated from the rest of the docstring by a blank line. The summary line may be on the same line as the opening quotes or on the next line. The entire docstring is indented the same as the quotes at its first line (see example below).
| A string in ['train', 'val', 'valid', 'test'], corresponding to | ||
| the set to be returned. | ||
| split: float32 | ||
| A float indicanting the fraction of data to be used for training. The |
| A float indicanting the fraction of data to be used for training. The | ||
| remaining data will be used for validation. | ||
| smooth_or_raw: string | ||
| A string in ['raw', 'smooth', 'both'], correspoding to the channels to be |
| remaining data will be used for validation. | ||
| smooth_or_raw: string | ||
| A string in ['raw', 'smooth', 'both'], correspoding to the channels to be | ||
| returned. Raw are profiles sampled from the raw image, smooth profiles |
There was a problem hiding this comment.
Raw --> 'raw'
smooth --> 'smooth'
| import numpy as np | ||
| from PIL import Image | ||
| import re | ||
| import warnings |
There was a problem hiding this comment.
Remove unused imports. Please run a PEP8 check on the whole file and fix all the errors.
| elif self.smooth_raw_both == 'both': | ||
| ret['data'] = np.concatenate([smooth,raw],axis=2) | ||
|
|
||
| ret['data'] = np.expand_dims(ret['data'],axis=2) |
There was a problem hiding this comment.
I'd rather change the assert in the parallel_loaders and leave it 3D here.
We could add a is_1D=True attribute to the class (i.e., before the def __init__) and act accordingly in the parallel_loaders here. If you check for getattr(self, 'is_1D') it should be easy to change the asserts. You can also skip the random_transform in that case, cause all the transforms are for 2D data IIRC.
| return | ||
|
|
||
| def distinct_colours(self, n): | ||
| #returns n distinct colours using nipy_spectral colourmap |
There was a problem hiding this comment.
Indentation is not a multiple of 4.
|
|
||
| if self.set_has_GT and self._void_labels != []: | ||
| if self.set_has_GT and self._void_labels != [] or \ | ||
| max(self._cmap.keys()) > self.non_void_nclasses-1: |
There was a problem hiding this comment.
self._cmap is an optional attribute, this will not work whenever it's not set. Also I don't see how the rest of the code can work if self.set_has_GT==False or if self._void_labels == []. What is the condition that you are trying to check here?
| seq_x = seq_x.squeeze(2) | ||
| seq_y = seq_y.squeeze(2) if self.return_one_hot else \ | ||
| seq_y.squeeze(1) | ||
| raw_data = raw_data.squeeze(2) |
There was a problem hiding this comment.
If you modify the parallel_loaders to accept 3D data out of load_sequence you shouldn't need all these modifications.
|
@pep8speaks suggest diff |
|
Here you go with the gist !
|
|
Hey Adri! Do you have time to close this PR? |
3 participants
This is the dataset that @lamblin mentioned. I merged the changes to the parallel_loader.py, so that we only have one file. This includes the dataset file cortical_layers.py (for any number of layers) and the parallel_loader.py small modifications to handle the 1D cases.