Add visual prompting task#2203

Merged

wonjuleee merged 136 commits intoopen-edge-platform:developfrom

JihwanEom:jeom/visual-prompting

Jul 3, 2023

Commits on Jul 3, 2023

Enable visual prompting

JihwanEom
authored and
sungchul1
committed
Resolve bugs

JihwanEom
authored and
sungchul1
committed
Bugfix

JihwanEom
authored and
sungchul1
committed
Performance check

JihwanEom
authored and
sungchul1
committed
Handle no objects in image

JihwanEom
authored and
sungchul1
committed
Update

JihwanEom
authored and
sungchul1
committed
Add SAM modules to avoid importing api
sungchul1
committed
(WIP) Add OTXVisualPromptingTask & related modules
sungchul1
committed
(WIP) enable _init_task in `PytorchLightningVisualPromptingTask`
sungchul1
committed
(WIP) enable `configure` in `PytorchLightningVisualPromptingTask`
sungchul1
committed
Add configuration.yaml & fix typo in template.yaml

JihwanEom
authored and
sungchul1
committed
Remove wip tasks to fit anomalib constructure
sungchul1
committed
(WIP) Dataset refactoring
sungchul1
committed
Set SAM pl module
sungchul1
committed
(WIP) Refactoring tasks
sungchul1
committed
(WIP) Set configs
sungchul1
committed
Refine Dataset (to be refactored)
sungchul1
committed
Enable to adapt configs & load model
sungchul1
committed
fix indent
sungchul1
committed
move setting transform into setup of datamodule
sungchul1
committed
Set train task & don't use project
sungchul1
committed
(WIP) Set `otx eval`
sungchul1
committed
fix
sungchul1
committed
Update configs
sungchul1
committed
Fix
sungchul1
committed
Split backbone from image encoder
sungchul1
committed
Fix
sungchul1
committed
Update callbacks & configs
sungchul1
committed
Add early stopping & increase max epochs
sungchul1
committed
(WIP) Add dataset adapter
sungchul1
committed
Enable VisualPromptingDatasetAdapter
sungchul1
committed
Use dataset config in dataset and datamodule
sungchul1
committed
(WIP) logic to generate masks and bboxes
sungchul1
committed
Revert & fix some changes
sungchul1
committed
Enable dataset
sungchul1
committed
Update modules related to configs
sungchul1
committed
warn -> warning
sungchul1
committed
Pass conversion error
sungchul1
committed
Update logs using torchmetrics
sungchul1
committed
Add loss type & refactoring
sungchul1
committed
(WIP) Enable otx eval
sungchul1
committed
Add Dice & fix when max_epochs=0
sungchul1
committed
Enable to change use_mask in cli
sungchul1
committed
Update checkpoint path name
sungchul1
committed
Enable to save configs using `save_hyperparameters`
sungchul1
committed
(WIP) Enable otx eval 2
sungchul1
committed
Fix
sungchul1
committed
Fix not to set label in inference callback
sungchul1
committed
Check not implemented modules
sungchul1
committed
Update adding performance
sungchul1
committed
Fix to consider padding when eval
sungchul1
committed
Enable to handle voc and common_semantic_segmentation format following semantic segmentation dataset adapter
sungchul1
committed
Set `use_mask` to False as default to match previous experiments
sungchul1
committed
Enable to use voc and custom_semantic_segmentation format
sungchul1
committed
Edit for when model_ckpt is empty str
sungchul1
committed
Edit to use use_mask instead of config in dataset
sungchul1
committed
Update to convert label when using mask on common_semantic_segmentation
sungchul1
committed
Add unit test for visual prompting dataset adapter
sungchul1
committed
Add common_semantic_segmentation data format in config manager
sungchul1
committed
Add visual prompting task in test_model_template
sungchul1
committed
Move backbone -> backbones
sungchul1
committed
Add condition for when checkpoint is None
sungchul1
committed
Add unit test for segment anything
sungchul1
committed
Add unit test for sam image encoder
sungchul1
committed
Add unit test for sam prompt encoder
sungchul1
committed
Use absolute import
sungchul1
committed
Add unit test for sam mask decoder
sungchul1
committed
Add unit test for vit backbone
sungchul1
committed
Fix
sungchul1
committed
Add unit test for dataset
sungchul1
committed
Remove unnecessary torch.stack
sungchul1
committed
Add unit test for transforms
sungchul1
committed
Fix
sungchul1
committed
Fix
sungchul1
committed
Update adding mask as annotation
sungchul1
committed
Add helper for visual prompting
sungchul1
committed
Update test_dataset
sungchul1
committed
Fix
sungchul1
committed
Add unit tests for inference callback
sungchul1
committed
Update Mock config
sungchul1
committed
Add unit tests of train task
sungchul1
committed
Move config & add output_path in config
sungchul1
committed
Add unit tests for config
sungchul1
committed
Add unit tests for inference task
sungchul1
committed
Remove unnecessary block
sungchul1
committed
Enable to update use_mask into data.yaml
sungchul1
committed
Remove use_mask in dataset and config
sungchul1
committed
Remove unnecessary part
sungchul1
committed
Update config part
sungchul1
committed
Apply updating data_yaml on only train
sungchul1
committed
output_path -> base_dir
sungchul1
committed
Update to get use_mask from otx_dataset
sungchul1
committed
Move test_visual_prompting_config
sungchul1
committed
Remove confused import
sungchul1
committed
Update use_mask in unit test
sungchul1
committed
Fix
sungchul1
committed
Add visual prompting & whem mode=eval
sungchul1
committed
Docstring
sungchul1
committed
Add max_epochs in configuration.yaml
sungchul1
committed
pylint
sungchul1
committed
Add integration test
sungchul1
committed
Add flag
sungchul1
committed
Update changelog
sungchul1
committed
Fix
sungchul1
committed
Fix (when using tox, val_dataset wasn't loaded)
sungchul1
committed
Fix
sungchul1
committed
pylint
sungchul1
committed
Fix file name
sungchul1
committed
Add __init__
sungchul1
committed
Add __init__ to import ov modules
sungchul1
committed
Resolve error about different data_root structure
sungchul1
committed
Resolve error about #domains
sungchul1
committed
Enable flag to integration and unit
sungchul1
committed
Edit for tasks have other task_type format
sungchul1
committed
Update otx/algorithms/visual_prompting/adapters/pytorch_lightning/config/visual_prompting_config.py

sungchul1
and
eunwoosh
committed
Fix type hint
sungchul1
committed
Update print -> logger
sungchul1
committed
Enable to load & resume checkpoints
sungchul1
committed
Fix
sungchul1
committed
Fix
sungchul1
committed
Fix for eval
sungchul1
committed
Fix
sungchul1
committed
Update docstring
sungchul1
committed
Update template
sungchul1
committed
pylint
sungchul1
committed
Divide transforms into sam's and others
sungchul1
committed
Update copyright
sungchul1
committed
Update condition
sungchul1
committed
Fix
sungchul1
committed
Update path
sungchul1
committed
Update variable name
sungchul1
committed
Fix #iou_predictions
sungchul1
committed
Makes black happy
sungchul1
committed
Fix typo
sungchul1
committed
Fix shuffle in train dataloader
sungchul1
committed
Fix #loaded datasets
sungchul1
committed