Repository navigation
Schirrmeister2017 High-Gamma Dataset from EDF - #265
sylvchev merged 3 commits into
Conversation
Sorry, something went wrong.
| ] | ||
| return cnt | ||
| sessions = { | ||
| "session_1": {"train": train_raw, "test": test_raw}, |
There was a problem hiding this comment.
| "session_1": {"train": train_raw, "test": test_raw}, | |
| "session_0": {"run_0": train_raw, "run_1": test_raw}, |
We are trying to use consistent name for session and run number, i.e. using integer and starting from 0. Is that ok to rename the dict file as suggested?
Sorry, something went wrong.
There was a problem hiding this comment.
I think could be nice to keep it train/test to keep backwards compatibility with code that did it this way? Also for this dataset, train was not one run, but typically 11 consecutive runs concatenated, and test 2 such consecutive runs. What do you think?
Sorry, something went wrong.
There was a problem hiding this comment.
I agree that in this case, we could keep the name "train" and "test". If we need to access specific run in the future, we could open a PR for adding a parameter to the dataset class to specify that we want to load each run (without train/test name).
So, let's merge!
Sorry, something went wrong.
|
great thanks! do you already have a plan for when your next PyPI release will be? |
Sorry, something went wrong.
|
I'll check with Jan, as he is currently adding new datasets, but I think we could makea release in a few days. |
Sorry, something went wrong.
* Set download dir test and example (#249) * Update to dataset_search call in FilterBank Motor Imagery * Removing completed #fixme * Removing total_classes argument from dataset_search call in FilterBank MI This was earlier deprecated in 55f77ae * set_download_dir test and example * adding pre-commit modifications * Update whats_new.rst * Update examples/changing_download_directory.py Co-authored-by: Sylvain Chevallier <sylvain.chevallier@uvsq.fr> * Update examples/changing_download_directory.py Co-authored-by: Sylvain Chevallier <sylvain.chevallier@uvsq.fr> * Bump pillow from 8.4.0 to 9.0.0 (#253) Bumps [pillow](https://github.com/python-pillow/Pillow) from 8.4.0 to 9.0.0. - [Release notes](https://github.com/python-pillow/Pillow/releases) - [Changelog](https://github.com/python-pillow/Pillow/blob/main/CHANGES.rst) - [Commits](python-pillow/Pillow@8.4.0...9.0.0) --- updated-dependencies: - dependency-name: pillow dependency-type: indirect ... Signed-off-by: dependabot[bot] <support@github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> * Fix Schirrmeister2017 error (#255) * correct event loading error, renaming session and runs * add whats new * Removing dependency of Physionet MI download on mne method (#257) * Update physionet_mi.py * consistency of runs numbering * Update whats_new.rst * f-string edits Co-authored-by: Sylvain Chevallier <sylvain.chevallier@uvsq.fr> * f-string edits Co-authored-by: Sylvain Chevallier <sylvain.chevallier@uvsq.fr> Co-authored-by: Sylvain Chevallier <sylvain.chevallier@uvsq.fr> * Correct MAMEM issues (#256) * switch mamem session to runs, use predictable names * update docstring in evaluation, for building documentation * update Lee2017 docstring for correct documentation. * update whats new * switch SSVEP example to within session * correct typo and rebase * correct typos on examples * Progress bars (#258) * Progress bars for downloads using pooch functionality * Rectification of f-string in PhysionetMI * Evaluations subject level progress bar CV test subject level in the case of CrossSubjectEvaluation * Update poetry.lock * Update pyproject.toml * dependencies * Apply suggestions from code review (mne.utils to tqdm direct) Co-authored-by: Sylvain Chevallier <sylvain.chevallier@uvsq.fr> * Update poetry.lock * tqdm arg * Update whats_new.rst * Update mistune dep Co-authored-by: Sylvain Chevallier <sylvain.chevallier@uvsq.fr> * fix doc url in readme (#262) * fix doc url in readme * correct links in the docs * Schirrmeister2017 High-Gamma Dataset from EDF (#265) * loading Schirrmeister2017 High-Gamma Dataset from EDF * remove commented import of requests module * rename to session_0 * added 13 + 12 subjects speller datasets by huebner (#260) * added 13 + 12 subjects speller datasets by huebner * clean up legacy run splitting code * added use_blocks_as_sessions parameter for data Co-authored-by: Sylvain Chevallier <sylvain.chevallier@uvsq.fr> * added Spot Auditory oddball dataset (#266) * added Spot Auditory oddball dataset * replaced usage of deprecated dl.data_path Co-authored-by: Sylvain Chevallier <sylvain.chevallier@uvsq.fr> * Visualize all ERP datasets (#261) * Visualize all ERP datasets * * use paradigm.datasets instead of manual list * more verbose sanity check script * fix epo data leak + remove title bf * moved data visualization added disclaimer regarding data size Co-authored-by: Sylvain Chevallier <sylvain.chevallier@uvsq.fr> * update to v0.4.5 (#269) * update to v0.4.5 * update poetry and requirements * correct pre-commit error and add code coverage (#271) Co-authored-by: Divyesh Narayanan <47829318+Div12345@users.noreply.github.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: robintibor <robintibor@gmail.com> Co-authored-by: Jan Sosulski <mail@jan-sosulski.de>
|
The 0.4.5 version is now on pypi, integrating this PR to update your dataset. Is that working on your side @robintibor ? |
Sorry, something went wrong.
|
Yeah it's working thanks a lot! :) |
Sorry, something went wrong.
We decided to reupload our dataset in EDF to have a more common format. This code will load the edf versions into moabb. There can be minimal differences in the loaded signal values to before, but fairly extensive decoding experiments with our deep networks show accuracies are only minimally affected at most.
Would be great to have it merged.