Conversation
crangelsmith
changed the title
…f parameters if pose = 0 , #290
…ta files loaded are not eval files
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
…ormat when functions are called in the postprocess notebook. + organising state files by date for restart flag
crangelsmith
changed the title
marjanfamili
left a comment
marjanfamili
left a comment
There was a problem hiding this comment.
The code runs fine, however I find all the new additions slightly confusing, specially because one has to switch back and force between caked and affinity to see what all the functions are. it would be good to have some short comments to explain what is happening when they are being used.
I have tested this on Mac for Mnist data with few transformations. The installation of caked works well in existing python environment without any conflict.
| x = self.voxel_transformation(data) | ||
| # read data from path | ||
| data = np.array(self.read(self.paths[index])) | ||
| x = self.transformation(data) |
There was a problem hiding this comment.
what is self.transformation() ?
There was a problem hiding this comment.
this is the equivalent to voxel_transformation, but as we should be able to deal with very different datasets (1D, 2D and 3D) it was more clear to leave it a transformation only.
Done in #88bbd47 |
Added more comments for context in 88bbd47 |
2 participants
Caked is being developed to allow for a more flexible data loader. It should do everything that is already implemented in affinity and more (e.g integration with CZI portal, parakeet, etc).
This PR, integrates Caked with affinity, making the
data.pya lot lighter in code.Commit history is a mess due to me trying to keep up to date with
develop, but the files changed are the right ones to review.