Mergeback 1.9.0 to develop (openvinotoolkit#1604)

<!-- Contributing guide:
https://github.com/openvinotoolkit/datumaro/blob/develop/CONTRIBUTING.md
-->

### Summary

<!--
Resolves openvinotoolkit#111 and openvinotoolkit#222.
Depends on openvinotoolkit#1000 (for series of dependent commits).

This PR introduces this capability to make the project better in this
and that.

- Added this feature
- Removed that feature
- Fixed the problem openvinotoolkit#1234
-->

### How to test
<!-- Describe the testing procedure for reviewers, if changes are
not fully covered by unit tests or manual testing can be complicated.
-->

### Checklist
<!-- Put an 'x' in all the boxes that apply -->
- [ ] I have added unit tests to cover my changes.​
- [ ] I have added integration tests to cover my changes.​
- [ ] I have added the description of my changes into
[CHANGELOG](https://github.com/openvinotoolkit/datumaro/blob/develop/CHANGELOG.md).​
- [ ] I have updated the
[documentation](https://github.com/openvinotoolkit/datumaro/tree/develop/docs)
accordingly

### License

- [ ] I submit _my code changes_ under the same [MIT
License](https://github.com/openvinotoolkit/datumaro/blob/develop/LICENSE)
that covers the project.
  Feel free to contact the maintainers if that's a concern.
- [ ] I have updated the license header for each file (see an example
below).

```python
# Copyright (C) 2024 Intel Corporation
#
# SPDX-License-Identifier: MIT
```

---------

Co-authored-by: Yunchu Lee <yunchu.lee@intel.com>
Co-authored-by: Wonju Lee <wonju.lee@intel.com>

CHANGELOG.md

@@ -9,6 +9,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### New features
 - Add a new CLI command: datum format
   (<https://github.com/openvinotoolkit/datumaro/pull/1570>)
+- Support language dataset for DmTorchDataset
+  (<https://github.com/openvinotoolkit/datumaro/pull/1592>)
 
 ### Enhancements
 - Change _Shape to Shape and add comments for subclasses of Shape

requirements-core.txt

@@ -64,3 +64,6 @@ json-stream
 
 # TabularValidator
 nltk
+
+# torch converter for language
+portalocker

setup.py

@@ -85,7 +85,7 @@ def parse_requirements(filename=CORE_REQUIREMENTS_FILE):
     extras_require={
         "tf": ["tensorflow"],
         "tfds": ["tensorflow-datasets<4.9.3"],
-        "torch": ["torch", "torchvision"],
+        "torch": ["torch", "torchvision", "torchtext==0.16.0"],
         "default": DEFAULT_REQUIREMENTS,
     },
     ext_modules=ext_modules,

src/datumaro/plugins/framework_converter.py

@@ -1,4 +1,4 @@
-# Copyright (C) 2023 Intel Corporation
+# Copyright (C) 2023-2024 Intel Corporation
 #
 # SPDX-License-Identifier: MIT
 
@@ -17,6 +17,7 @@
     "detection": AnnotationType.bbox,
     "instance_segmentation": AnnotationType.polygon,
     "semantic_segmentation": AnnotationType.mask,
+    "tabular": [AnnotationType.label, AnnotationType.caption],
 }
 
 
@@ -88,7 +89,10 @@ def _gen_item(self, idx: int):
                 if ann.type == TASK_ANN_TYPE[self.task]
             ]
             label = mask_tools.merge_masks((mask, label_id) for mask, label_id in masks)
-
+        elif self.task == "tabular":
+            label = [
+                ann.as_dict() for ann in item.annotations if ann.type in TASK_ANN_TYPE[self.task]
+            ]
         return image, label
 
 
@@ -103,15 +107,58 @@ def __init__(
             task: str,
             transform: Optional[Callable] = None,
             target_transform: Optional[Callable] = None,
+            target: Optional[str] = None,
+            tokenizer: Optional[tuple[Callable, Callable]] = None,
+            vocab: Optional[tuple[Callable, Callable]] = None,
         ):
             super().__init__(dataset=dataset, subset=subset, task=task)
 
             self.transform = transform
             self.target_transform = target_transform
 
+            if self.task == "tabular":
+                if not isinstance(target, dict):
+                    raise ValueError(
+                        "Target should be a dictionary with 'input' and 'output' keys."
+                    )
+                self.input_target = target.get("input")
+                self.output_target = target.get("output")
+                if not self.input_target:
+                    raise ValueError(
+                        "Please provide target column for tabular task which is used for input"
+                    )
+
+                if not (tokenizer and vocab):
+                    raise ValueError("Both tokenizer and vocab must be provided for tabular task")
+                self.tokenizer = tokenizer
+                self.vocab = vocab
+
         def __getitem__(self, idx):
             image, label = self._gen_item(idx)
 
+            if self.task == "tabular":
+                text = image()[self.input_target]
+
+                if self.output_target:
+                    src_tokenizer, tgt_tokenizer = self.tokenizer
+                    src_vocab, tgt_vocab = self.vocab
+                    src_tokens = src_tokenizer(text)
+                    src_token_ids = src_vocab(src_tokens)
+
+                    label_text = label[0]["caption"].split(f"{self.output_target}:")[-1]
+                    tgt_tokens = tgt_tokenizer(label_text)
+                    tgt_token_ids = tgt_vocab(tgt_tokens)
+
+                    return torch.tensor(src_token_ids, dtype=torch.long), torch.tensor(
+                        tgt_token_ids, dtype=torch.long
+                    )
+                else:
+                    tokens = self.tokenizer(text)
+                    token_ids = self.vocab(tokens)
+                    return torch.tensor(token_ids, dtype=torch.long), torch.tensor(
+                        label[0]["label"], dtype=torch.long
+                    )
+
             if len(image.shape) == 2:
                 image = np.expand_dims(image, axis=-1)