autogluon.multimodal 1.1.2b20241105__tar.gz → 1.1.2b20241107__tar.gz

{autogluon.multimodal-1.1.2b20241105 → autogluon.multimodal-1.1.2b20241107}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: autogluon.multimodal
-Version: 1.1.2b20241105
+Version: 1.1.2b20241107
 Summary: Fast and Accurate ML in 3 Lines of Code
 Home-page: https://github.com/autogluon/autogluon
 Author: AutoGluon Community

{autogluon.multimodal-1.1.2b20241105 → autogluon.multimodal-1.1.2b20241107}/src/autogluon/multimodal/learners/object_detection.py

@@ -74,6 +74,8 @@ class ObjectDetectionLearner(BaseLearner):
         )
         check_if_packages_installed(problem_type=self._problem_type)
 
+        self._config = self.get_config_per_run(config=self._config, hyperparameters=hyperparameters)
+
         self._output_shape = num_classes
         self._classes = classes
         self._sample_data_path = sample_data_path
@@ -102,10 +104,18 @@ class ObjectDetectionLearner(BaseLearner):
     def setup_detection_train_tuning_data(self, max_num_tuning_data, seed, train_data, tuning_data):
         if isinstance(train_data, str):
             self._detection_anno_train = train_data
-            train_data = from_coco_or_voc(train_data, "train")  # TODO: Refactor to use convert_data_to_df
+            train_data = from_coco_or_voc(
+                train_data,
+                "train",
+                coco_root=self._config.model.mmdet_image.coco_root,
+            )  # TODO: Refactor to use convert_data_to_df
             if tuning_data is not None:
                 self.detection_anno_train = tuning_data
-                tuning_data = from_coco_or_voc(tuning_data, "val")  # TODO: Refactor to use convert_data_to_df
+                tuning_data = from_coco_or_voc(
+                    tuning_data,
+                    "val",
+                    coco_root=self._config.model.mmdet_image.coco_root,
+                )  # TODO: Refactor to use convert_data_to_df
                 if max_num_tuning_data is not None:
                     if len(tuning_data) > max_num_tuning_data:
                         tuning_data = tuning_data.sample(
@@ -114,10 +124,16 @@ class ObjectDetectionLearner(BaseLearner):
         elif isinstance(train_data, pd.DataFrame):
             self._detection_anno_train = None
             # sanity check dataframe columns
-            train_data = object_detection_data_to_df(train_data)
+            train_data = object_detection_data_to_df(
+                train_data,
+                coco_root=self._config.model.mmdet_image.coco_root,
+            )
             if tuning_data is not None:
                 self.detection_anno_train = tuning_data
-                tuning_data = object_detection_data_to_df(tuning_data)
+                tuning_data = object_detection_data_to_df(
+                    tuning_data,
+                    coco_root=self._config.model.mmdet_image.coco_root,
+                )
                 if max_num_tuning_data is not None:
                     if len(tuning_data) > max_num_tuning_data:
                         tuning_data = tuning_data.sample(
@@ -556,7 +572,9 @@ class ObjectDetectionLearner(BaseLearner):
         if isinstance(anno_file_or_df, str):
             anno_file = anno_file_or_df
             data = from_coco_or_voc(
-                anno_file, "test"
+                anno_file,
+                "test",
+                coco_root=self._config.model.mmdet_image.coco_root,
             )  # TODO: maybe remove default splits hardcoding (only used in VOC)
             if os.path.isdir(anno_file):
                 eval_tool = "torchmetrics"  # we can only use torchmetrics for VOC format evaluation.
@@ -636,7 +654,10 @@ class ObjectDetectionLearner(BaseLearner):
                 eval_tool=eval_tool,
             )
         else:
-            data = object_detection_data_to_df(data)
+            data = object_detection_data_to_df(
+                data,
+                coco_root=self._config.model.mmdet_image.coco_root,
+            )
             return self.evaluate_coco(
                 anno_file_or_df=data,
                 metrics=metrics,
@@ -676,7 +697,10 @@ class ObjectDetectionLearner(BaseLearner):
         self.ensure_predict_ready()
         ret_type = BBOX
         if self._problem_type == OBJECT_DETECTION:
-            data = object_detection_data_to_df(data)
+            data = object_detection_data_to_df(
+                data,
+                coco_root=self._config.model.mmdet_image.coco_root,
+            )
             if self._label_column not in data:
                 self._label_column = None
 

{autogluon.multimodal-1.1.2b20241105 → autogluon.multimodal-1.1.2b20241107}/src/autogluon/multimodal/utils/object_detection.py

@@ -145,7 +145,9 @@ def object_detection_df_to_coco(data: pd.DataFrame, save_path: Optional[str] = N
     return output_json_dict
 
 
-def object_detection_data_to_df(data: Union[pd.DataFrame, dict, list, str]) -> pd.DataFrame:
+def object_detection_data_to_df(
+    data: Union[pd.DataFrame, dict, list, str], coco_root: Optional[str] = None
+) -> pd.DataFrame:
     """
     Construct a dataframe from a data dictionary, json file path (for COCO), folder path (for VOC),
     image path (for single image), list of image paths (for multiple images)
@@ -163,7 +165,7 @@ def object_detection_data_to_df(data: Union[pd.DataFrame, dict, list, str]) -> p
         return from_list(data)
     if isinstance(data, str):
         if os.path.isdir(data) or data.endswith(".json"):
-            return from_coco_or_voc(data)
+            return from_coco_or_voc(data, coco_root=coco_root)
         return from_str(data)
     if isinstance(data, pd.DataFrame):
         sanity_check_dataframe(data)
@@ -700,7 +702,7 @@ def _check_load_coco_bbox(
 
 def from_coco(
     anno_file: Optional[str],
-    root: Optional[str] = None,
+    coco_root: Optional[str] = None,
     min_object_area: Optional[Union[int, float]] = 0,
     use_crowd: Optional[bool] = False,
 ):
@@ -718,7 +720,7 @@ def from_coco(
     ----------
     anno_file
         The path to the annotation file.
-    root
+    coco_root
         Root of the COCO folder. The default relative root folder (if set to `None`) is `anno_file/../`.
     min_object_area
         Minimum object area to consider.
@@ -740,16 +742,19 @@ def from_coco(
     coco = COCO(anno_file)
 
     # get data root
-    if isinstance(root, Path):
-        root = str(root.expanduser().resolve())
-    elif isinstance(root, str):
-        root = os.path.abspath(os.path.expanduser(root))
-    elif root is None:
+    if isinstance(coco_root, Path):
+        coco_root = str(coco_root.expanduser().resolve())
+    elif isinstance(coco_root, str):
+        coco_root = os.path.abspath(os.path.expanduser(coco_root))
+    elif coco_root is None:
         # try to use the default coco structure
-        root = os.path.join(os.path.dirname(anno_file), "..")
-        logger.info(f"Using default root folder: {root}. Specify `root=...` if you feel it is wrong...")
+        coco_root = os.path.join(os.path.dirname(anno_file), "..")
+        logger.info(
+            f"Using default root folder: {coco_root}. "
+            "Specify `model.mmdet_image.coco_root=...` in hyperparameters if you think it is wrong."
+        )
     else:
-        raise ValueError("Unable to parse root: {}".format(root))
+        raise ValueError("Unable to parse coco_root: {}".format(coco_root))
 
     # support prediction using data with no annotations
     # note that data with annotation can be used for prediction without any changes
@@ -764,9 +769,9 @@ def from_coco(
     for entry in coco.loadImgs(image_ids):
         if "coco_url" in entry:
             dirname, filename = entry["coco_url"].split("/")[-2:]
-            abs_path = os.path.join(root, dirname, filename)
+            abs_path = os.path.join(coco_root, dirname, filename)
         else:
-            abs_path = os.path.join(root, entry["file_name"])
+            abs_path = os.path.join(coco_root, entry["file_name"])
         if not os.path.exists(abs_path):
             logger.warning(f"File skipped since not exists: {abs_path}.")
             continue
@@ -1146,7 +1151,7 @@ def process_voc_annotations(
         f.writelines("\n".join(xml_file_names))
 
 
-def from_coco_or_voc(file_path: str, splits: Optional[Union[str]] = None):
+def from_coco_or_voc(file_path: str, splits: Optional[Union[str]] = None, coco_root: Optional[str] = None):
     """
     Convert the data from coco or voc format to pandas Dataframe.
 
@@ -1167,7 +1172,7 @@ def from_coco_or_voc(file_path: str, splits: Optional[Union[str]] = None):
         # VOC use dir as input
         return from_voc(root=file_path, splits=splits)
     else:
-        return from_coco(file_path)
+        return from_coco(file_path, coco_root=coco_root)
 
 
 def get_coco_format_classes(sample_data_path: str):
@@ -1539,11 +1544,13 @@ def save_result_df(
     return result_df
 
 
-def save_result_coco_format(detection_data_path, pred, result_path):
+def save_result_coco_format(detection_data_path, pred, result_path, coco_root: Optional[str] = None):
     coco_dataset = COCODataset(detection_data_path)
     result_name, _ = os.path.splitext(result_path)
     result_path = result_name + ".json"
-    coco_dataset.save_result(pred, from_coco_or_voc(detection_data_path, "test"), save_path=result_path)
+    coco_dataset.save_result(
+        pred, from_coco_or_voc(detection_data_path, "test", coco_root=coco_root), save_path=result_path
+    )
     logger.info(25, f"Saved detection result to {result_path}")
 
 

{autogluon.multimodal-1.1.2b20241105 → autogluon.multimodal-1.1.2b20241107}/src/autogluon/multimodal/version.py

@@ -1,3 +1,3 @@
 """This is the autogluon version file."""
-__version__ = '1.1.2b20241105'
+__version__ = '1.1.2b20241107'
 __lite__ = False

{autogluon.multimodal-1.1.2b20241105 → autogluon.multimodal-1.1.2b20241107}/src/autogluon.multimodal.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: autogluon.multimodal
-Version: 1.1.2b20241105
+Version: 1.1.2b20241107
 Summary: Fast and Accurate ML in 3 Lines of Code
 Home-page: https://github.com/autogluon/autogluon
 Author: AutoGluon Community

{autogluon.multimodal-1.1.2b20241105 → autogluon.multimodal-1.1.2b20241107}/src/autogluon.multimodal.egg-info/requires.txt

@@ -5,10 +5,10 @@ scikit-learn<1.5.3,>=1.4.0
 Pillow<12,>=10.0.1
 tqdm<5,>=4.38
 boto3<2,>=1.10
-torch<2.5,>=2.2
-lightning<2.4,>=2.2
-transformers[sentencepiece]<4.41.0,>=4.38.0
-accelerate<0.34.0,>=0.21.0
+torch<2.6,>=2.2
+lightning<2.6,>=2.2
+transformers[sentencepiece]<5,>=4.38.0
+accelerate<1.0,>=0.32.0
 requests<3,>=2.30
 jsonschema<4.22,>=4.18
 seqeval<1.3.0,>=1.2.2
@@ -20,9 +20,9 @@ text-unidecode<1.4,>=1.3
 torchmetrics<1.3.0,>=1.2.0
 nptyping<2.5.0,>=1.4.4
 omegaconf<2.3.0,>=2.1.1
-autogluon.core[raytune]==1.1.2b20241105
-autogluon.features==1.1.2b20241105
-autogluon.common==1.1.2b20241105
+autogluon.core[raytune]==1.1.2b20241107
+autogluon.features==1.1.2b20241107
+autogluon.common==1.1.2b20241107
 pytorch-metric-learning<2.4,>=1.3.0
 nlpaug<1.2.0,>=1.1.10
 nltk<3.9,>=3.4.5