[Dataset] Add VisA dataset #824
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
* move sample generation to datamodule instead of dataset * move sample generation from init to setup * remove inference stage and add base classes * replace dataset classes with AnomalibDataset * move setup to base class, create samples as class method * update docstrings * refactor btech to new format * allow training with no anomalous data * remove MVTec name from comment * raise NotImplementedError in base class * allow both png and bmp images for btech * use label_index to check if dataset contains anomalous images * refactor getitem in dataset class * use iloc for indexing * move dataloader getters to base class * refactor to add validate stage in setup * implement alternative datamodules solution * small improvements * improve design * remove unused constructor arguments * adapt btech to new design * add prepare_data method for mvtec * implement more generic random splitting function * update docstrings for folder module * ensure type consistency when performing operations on dataset * change imports * change variable names * replace pass with NotImplementedError * allow training on folder without test images * use relative path for normal_test_dir * fix dataset tests * update validation set parameter in configs * change default argument * use setter for samples * hint options for val_split_mode * update assert message and docstring * revert name change dataset vs datamodule * typing and docstrings * remove samples argument from dataset constructor * val/test -> eval * remove Split.Full from enum * sort samples when setting * update warn message * formatting * use setter when creating samples in dataset classes * add tests for new dataset class * add test case for label aware random split * update parameter name in inferencers * move _setup implementation to base class * address codacy issues * fix pylint issues * codacy * update example dataset config in docs * fix test * move base classes to separate files (avoid circular import) * add base classes * update docstring * fix imports * validation_split_mode -> val_split_mode * update docs * Update anomalib/data/base/dataset.py Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com> * get length from self.samples * assert unique indices * check is_setup for individual datasets Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com> * remove assert in __getitem_\ Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com> * Update anomalib/data/btech.py Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com> * clearer assert message * clarify list inversion in comment * comments and typing * validate contents of samples dataframe before setting * add file paths check * add seed to random_split function * fix expected columns * fix typo * add seed parameter to datamodules * set global seed in test entrypoint * add NONE option to valsplitmode * clarify setup behaviour in docstring * fix typo Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com> Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com>
* move sample generation to datamodule instead of dataset * move sample generation from init to setup * remove inference stage and add base classes * replace dataset classes with AnomalibDataset * move setup to base class, create samples as class method * update docstrings * refactor btech to new format * allow training with no anomalous data * remove MVTec name from comment * raise NotImplementedError in base class * allow both png and bmp images for btech * use label_index to check if dataset contains anomalous images * refactor getitem in dataset class * use iloc for indexing * move dataloader getters to base class * refactor to add validate stage in setup * implement alternative datamodules solution * small improvements * improve design * remove unused constructor arguments * adapt btech to new design * add prepare_data method for mvtec * implement more generic random splitting function * update docstrings for folder module * ensure type consistency when performing operations on dataset * change imports * change variable names * replace pass with NotImplementedError * allow training on folder without test images * use relative path for normal_test_dir * fix dataset tests * update validation set parameter in configs * change default argument * use setter for samples * hint options for val_split_mode * update assert message and docstring * revert name change dataset vs datamodule * typing and docstrings * remove samples argument from dataset constructor * val/test -> eval * remove Split.Full from enum * sort samples when setting * update warn message * formatting * use setter when creating samples in dataset classes * add tests for new dataset class * add test case for label aware random split * update parameter name in inferencers * move _setup implementation to base class * address codacy issues * fix pylint issues * codacy * update example dataset config in docs * fix test * move base classes to separate files (avoid circular import) * add base classes * update docstring * fix imports * validation_split_mode -> val_split_mode * update docs * Update anomalib/data/base/dataset.py Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com> * get length from self.samples * assert unique indices * check is_setup for individual datasets Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com> * remove assert in __getitem_\ Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com> * Update anomalib/data/btech.py Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com> * clearer assert message * clarify list inversion in comment * comments and typing * validate contents of samples dataframe before setting * add file paths check * add seed to random_split function * fix expected columns * fix typo * add pedestrian and avenue datasets and video utils * add seed parameter to datamodules * set global seed in test entrypoint * add NONE option to valsplitmode * clarify setup behaviour in docstring * add basic visualization for video datasets * simplify ucsdped implementation * add ucsd and avenue to __all__ * add default value for task * add tests for ucsd and avenue * add tests for video dataset and utils * add download info for avenue dataset * add download info for ucsd pedestrian dataset * more consistent naming * fix path to masks folder in gt dir * pass original image in batch to facilitate visualization * convert mask files for avenue * suppress warning due to torchvision bug * fix bug in avenue masks * store visualizations for each video in separate folder * rename parameters * add warning for clip_length > 1 * fix dataset tests * fix labels tensor shape bug * add pyav to requirements * add description for avenue dataset * use pathlib * Update anomalib/data/avenue.py Co-authored-by: Samet Akcay <samet.akcay@intel.com> * Update anomalib/data/avenue.py Co-authored-by: Samet Akcay <samet.akcay@intel.com> * Update anomalib/data/utils/video.py Co-authored-by: Samet Akcay <samet.akcay@intel.com> * Update anomalib/data/base/video.py Co-authored-by: Samet Akcay <samet.akcay@intel.com> * Update anomalib/data/base/video.py Co-authored-by: Samet Akcay <samet.akcay@intel.com> * Update anomalib/data/ucsd_ped.py Co-authored-by: Samet Akcay <samet.akcay@intel.com> * import video dataset from base * fix bug when collecting ucsd samples * clean up datamodules tests * fix tests * remove redundant test cases * retrieve masks as numpy array * use pathlib * variable name * pathlib * use preprocesser from arguments * fix indexing bug Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com> Co-authored-by: Samet Akcay <samet.akcay@intel.com>
* make val split ratio configurable * use DeprecationWarning, update config key
* add basic support for detection task * use enum for task type * formatting * small bugfix * add unit tests for bounding box conversion * update error message * use as_tensor * typing and docstring * explicit keyword arguments * simplify bbox handling in video dataset * docstring consistency * add missing licenses * add whitespace for readability * add missing license * Update anomalib/data/utils/boxes.py Co-authored-by: Samet Akcay <samet.akcay@intel.com> * Revert "Update anomalib/data/utils/boxes.py" This reverts commit cec6138. * add test case for custom collate function * docstring * add integration tests for detection dataloading * extend and clean up datamodules tests * add detection task type to visualizer tests * only show pred_boxes during inference * add detection support for torch inference * add detection support for openvino inference * test inference for all task types * pylint Co-authored-by: Samet Akcay <samet.akcay@intel.com>
* update deprecation messages * raise warnings as DeprecationWarning
* mask -> mask_dir * properly handle absolute and relative paths * make root path parameter optional * formatting * path -> root * update docs * remove options hint for name parameter * refactor function * Update anomalib/config/config.py Co-authored-by: Samet Akcay <samet.akcay@intel.com> * Update anomalib/config/config.py Co-authored-by: Samet Akcay <samet.akcay@intel.com> * make root and abnormal_dir optional * Update anomalib/data/folder.py Co-authored-by: Samet Akcay <samet.akcay@intel.com> Co-authored-by: Samet Akcay <samet.akcay@intel.com>
* move sample generation to datamodule instead of dataset * move sample generation from init to setup * remove inference stage and add base classes * replace dataset classes with AnomalibDataset * move setup to base class, create samples as class method * update docstrings * refactor btech to new format * allow training with no anomalous data * remove MVTec name from comment * raise NotImplementedError in base class * allow both png and bmp images for btech * use label_index to check if dataset contains anomalous images * refactor getitem in dataset class * use iloc for indexing * move dataloader getters to base class * refactor to add validate stage in setup * implement alternative datamodules solution * small improvements * improve design * remove unused constructor arguments * adapt btech to new design * add prepare_data method for mvtec * implement more generic random splitting function * update docstrings for folder module * ensure type consistency when performing operations on dataset * change imports * change variable names * replace pass with NotImplementedError * allow training on folder without test images * use relative path for normal_test_dir * fix dataset tests * update validation set parameter in configs * change default argument * use setter for samples * hint options for val_split_mode * update assert message and docstring * revert name change dataset vs datamodule * typing and docstrings * remove samples argument from dataset constructor * val/test -> eval * remove Split.Full from enum * sort samples when setting * update warn message * formatting * use setter when creating samples in dataset classes * add tests for new dataset class * add test case for label aware random split * update parameter name in inferencers * move _setup implementation to base class * address codacy issues * fix pylint issues * codacy * update example dataset config in docs * fix test * move base classes to separate files (avoid circular import) * add synthetic dataset class * move augmenter to data directory * add base classes * update docstring * use synthetic dataset in base datamodule * fix imports * clean up synthetic anomaly dataset implementation * fix mistake in augmenter * change default split ratio * remove accidentally added file * validation_split_mode -> val_split_mode * update docs * Update anomalib/data/base/dataset.py Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com> * get length from self.samples * assert unique indices * check is_setup for individual datasets Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com> * remove assert in __getitem_\ Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com> * Update anomalib/data/btech.py Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com> * clearer assert message * clarify list inversion in comment * comments and typing * validate contents of samples dataframe before setting * add file paths check * add seed to random_split function * fix expected columns * fix typo * add seed parameter to datamodules * set global seed in test entrypoint * add NONE option to valsplitmode * clarify setup behaviour in docstring * add logging message * use val_split_ratio for synthetic validation set * pathlib * make synthetic anomaly available for test set * update configs * add tests * simplify test set splitting logic * update docstring * add missing licence * split_normal_and_anomalous -> split_by_label * VideoAnomalib -> AnomalibVideo Co-authored-by: Joao P C Bertoldo <24547377+jpcbertoldo@users.noreply.github.com>
* properly handle NoneType mask_dir and add test case * fix wrong deprecation handling
* deprecate PreProcessor * update configs * update deprecation messages * update video dataset * update inference dataset * move transforms to data module * update and extend transform tests * fix cyclic import * add validity checks for image size and center crop * pass image size as tuple * update path to get_transforms * update error message * fix center crop tuple conversion * update inferencers * remove draem transform config * update changelog * fix cyclic import * add crop size vs image size check * improve readability * mypy * use enum to configure input normalization * update lightning inference * update inference dataset
handle empty box predictions
…e'` (#801) * enable none as split mode * use get to retrieve config keys * update deprecation message and config key
* apply pixel threshold to bbox detections * allow visualizing normal boxes * normalize box scores * fix bbox logic in base anomaly module * boxes_scores -> box_scores * fix inferencers
* infer box scores from anomaly maps * discard single pixel boxes * revert discard single pixel boxes * add test case for bbox scores * update torch inferencer * minor refactor
|
Thanks for integrating the dataset! Thoughts on adding a comparison of methods similar as done for MVTec in the Readme? Would be helpful for those in the community with little compute available :D |
|
Hi @ORippler, yeah, it's on our long to-do list. We hope to add it soon :) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
This PR adds the Visual Anomaly (VisA) dataset.
The dataset follows the same format as MVTec, so we could re-use the
make_mvtec_datasetfunction.The
make_mvtec_datasetfunction was slightly modified to make the mask file naming convention a bit more flexible (mvtec uses"000_mask.png", while visa uses"000.png").There was a lot of duplication in the download and extract functionality of the different datasets, so this was moved to a shared location.
Currently targeted to feature branch, but will re-target to main once #822 has been merged.
Some examples:
Known Issues
CI will probably fail, because the dataset is not yet installed on the CI machine.Changes
Checklist