paddlex 3.0.0rc0__py3-none-any.whl → 3.0.0rc1__py3-none-any.whl

paddlex/inference/models/image_unwarping/predictor.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -12,25 +12,21 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from typing import Any, Union, Dict, List, Tuple
+from typing import Any, Dict, List, Tuple, Union
+
 import numpy as np
 
 from ....modules.image_unwarping.model_list import MODELS
 from ...common.batch_sampler import ImageBatchSampler
 from ...common.reader import ReadImage
-from ..common import (
-    Normalize,
-    ToCHWImage,
-    ToBatch,
-    StaticInfer,
-)
-from ..base import BasicPredictor
+from ..base import BasePredictor
+from ..common import Normalize, ToBatch, ToCHWImage
 from .processors import DocTrPostProcess
 from .result import DocTrResult
 
 
-class WarpPredictor(BasicPredictor):
-    """WarpPredictor that inherits from BasicPredictor."""
+class WarpPredictor(BasePredictor):
+    """WarpPredictor that inherits from BasePredictor."""
 
     entities = MODELS
 
@@ -71,11 +67,7 @@ class WarpPredictor(BasicPredictor):
         preprocessors["ToCHW"] = ToCHWImage()
         preprocessors["ToBatch"] = ToBatch()
 
-        infer = StaticInfer(
-            model_dir=self.model_dir,
-            model_prefix=self.MODEL_FILE_PREFIX,
-            option=self.pp_option,
-        )
+        infer = self.create_static_infer()
 
         postprocessors = {"DocTrPostProcess": DocTrPostProcess()}
         return preprocessors, infer, postprocessors

paddlex/inference/models/image_unwarping/processors.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -12,10 +12,14 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+from typing import List, Tuple, Union
+
 import numpy as np
-from typing import List, Union, Tuple
+
+from ...utils.benchmark import benchmark
 
 
+@benchmark.timeit
 class DocTrPostProcess:
     """
     Post-processing class for cropping regions from images (though currently only performs scaling and color channel adjustments).

paddlex/inference/models/image_unwarping/result.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -13,8 +13,10 @@
 # limitations under the License.
 
 import copy
+
 import numpy as np
-from ...common.result import BaseCVResult, StrMixin, JsonMixin
+
+from ...common.result import BaseCVResult, JsonMixin
 
 
 class DocTrResult(BaseCVResult):

paddlex/inference/models/instance_segmentation/__init__.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

paddlex/inference/models/instance_segmentation/predictor.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -12,20 +12,16 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from typing import Any, Union, Dict, List, Tuple, Sequence, Optional
+from typing import Any, List, Optional, Sequence, Tuple
+
 import numpy as np
 
 from ....modules.instance_segmentation.model_list import MODELS
-from ...common.batch_sampler import ImageBatchSampler
-from ..common import StaticInfer
-from ..object_detection.processors import (
-    ReadImage,
-    ToBatch,
-)
-from .processors import InstanceSegPostProcess
+from ....utils import logging
 from ..object_detection import DetPredictor
+from ..object_detection.processors import ReadImage, ToBatch
+from .processors import InstanceSegPostProcess
 from .result import InstanceSegResult
-from ....utils import logging
 
 
 class InstanceSegPredictor(DetPredictor):
@@ -85,11 +81,7 @@ class InstanceSegPredictor(DetPredictor):
         pre_ops.append(self.build_to_batch())
 
         # build infer
-        infer = StaticInfer(
-            model_dir=self.model_dir,
-            model_prefix=self.MODEL_FILE_PREFIX,
-            option=self.pp_option,
-        )
+        infer = self.create_static_infer()
 
         # build postprocess op
         post_op = self.build_postprocess()

paddlex/inference/models/instance_segmentation/processors.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -12,15 +12,11 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import os
-from typing import List, Sequence, Tuple, Union, Optional
+from typing import List, Optional
 
-import numpy as np
-from ....utils import logging
+from ...utils.benchmark import benchmark
 from ..object_detection.processors import restructured_boxes
 
-import cv2
-
 
 def extract_masks_from_boxes(boxes, masks):
     """
@@ -40,6 +36,7 @@ def extract_masks_from_boxes(boxes, masks):
     return new_masks
 
 
+@benchmark.timeit
 class InstanceSegPostProcess(object):
     """Save Result Transform"""
 

paddlex/inference/models/instance_segmentation/result.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -12,24 +12,25 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import os
-import cv2
-import numpy as np
 import copy
-import PIL
-from PIL import Image, ImageDraw, ImageFont
 
-from ...utils.color_map import get_colormap, font_colormap
-from ...common.result import BaseCVResult, StrMixin, JsonMixin
-from ....utils.fonts import PINGFANG_FONT_FILE_PATH
+import numpy as np
+from PIL import Image
+
+from ....utils.deps import function_requires_deps, is_dep_available
+from ...common.result import BaseCVResult, JsonMixin
+from ...utils.color_map import get_colormap
 from ..object_detection.result import draw_box
 
+if is_dep_available("opencv-contrib-python"):
+    import cv2
+
 
+@function_requires_deps("opencv-contrib-python")
 def draw_segm(im, masks, mask_info, alpha=0.7):
     """
     Draw segmentation on image
     """
-    mask_color_id = 0
     w_ratio = 0.4
     color_list = get_colormap(rgb=True)
     im = np.array(im).astype("float32")

paddlex/inference/models/keypoint_detection/__init__.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

paddlex/inference/models/keypoint_detection/predictor.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -19,10 +19,9 @@ import numpy as np
 from ....modules.keypoint_detection.model_list import MODELS
 from ....utils import logging
 from ...common.batch_sampler import ImageBatchSampler
-
 from ..common import ToBatch
 from ..object_detection import DetPredictor
-from .processors import TopDownAffine, KptPostProcess
+from .processors import KptPostProcess, TopDownAffine
 from .result import KptResult
 
 

paddlex/inference/models/keypoint_detection/processors.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -13,14 +13,18 @@
 # limitations under the License.
 
 import math
-from typing import List, Sequence, Tuple, Union, Optional
+from typing import List, Optional, Sequence, Tuple, Union
 
-import cv2
 import numpy as np
 from numpy import ndarray
 
+from ....utils.deps import class_requires_deps, is_dep_available
+from ...utils.benchmark import benchmark
 from ..object_detection.processors import get_affine_transform
 
+if is_dep_available("opencv-contrib-python"):
+    import cv2
+
 Number = Union[int, float]
 Kpts = List[dict]
 
@@ -65,6 +69,8 @@ def get_warp_matrix(
     return matrix
 
 
+@benchmark.timeit
+@class_requires_deps("opencv-contrib-python")
 class TopDownAffine:
     """refer to https://github.com/open-mmlab/mmpose/blob/71ec36ebd63c475ab589afc817868e749a61491f/mmpose/datasets/transforms/topdown_transforms.py#L13
     Get the bbox image as the model input by affine transform.
@@ -196,6 +202,8 @@ def transform_preds(
     return target_coords
 
 
+@benchmark.timeit
+@class_requires_deps("opencv-contrib-python")
 class KptPostProcess:
     """Save Result Transform"""
 

paddlex/inference/models/keypoint_detection/result.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -12,16 +12,20 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import cv2
 import copy
 import math
-from PIL import Image
 
-import matplotlib.pyplot as plt
 import numpy as np
+from PIL import Image
 
+from ....utils.deps import function_requires_deps, is_dep_available
 from ...common.result import BaseCVResult, JsonMixin
 
+if is_dep_available("opencv-contrib-python"):
+    import cv2
+if is_dep_available("matplotlib"):
+    import matplotlib.pyplot as plt
+
 
 def get_color(idx):
     idx = idx * 3
@@ -29,6 +33,7 @@ def get_color(idx):
     return color
 
 
+@function_requires_deps("matplotlib", "opencv-contrib-python")
 def draw_keypoints(img, results, visual_thresh=0.1, ids=None):
     plt.switch_backend("agg")
     skeletons = results["keypoints"]

paddlex/inference/models/{3d_bev_detection → m_3d_bev_detection}/__init__.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.

paddlex/inference/models/{3d_bev_detection → m_3d_bev_detection}/predictor.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -12,38 +12,32 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from typing import Any, Union, Dict, List, Tuple, Iterator
 import shutil
 import tempfile
-from importlib import import_module
-import lazy_paddle
+from typing import Any, Dict, Iterator, List, Tuple
 
+from ....modules.m_3d_bev_detection.model_list import MODELS
 from ....utils import logging
 from ....utils.func_register import FuncRegister
-
-module_3d_bev_detection = import_module(".3d_bev_detection", "paddlex.modules")
-module_3d_model_list = getattr(module_3d_bev_detection, "model_list")
-MODELS = getattr(module_3d_model_list, "MODELS")
 from ...common.batch_sampler import Det3DBatchSampler
 from ...common.reader import ReadNuscenesData
-from ..common import StaticInfer
-from ..base import BasicPredictor
+from ..base import BasePredictor
 from ..base.predictor.base_predictor import PredictionWrap
 from .processors import (
+    GetInferInput,
+    LoadMultiViewImageFromFiles,
     LoadPointsFromFile,
     LoadPointsFromMultiSweeps,
-    LoadMultiViewImageFromFiles,
-    ResizeImage,
     NormalizeImage,
     PadImage,
+    ResizeImage,
     SampleFilterByKey,
-    GetInferInput,
 )
 from .result import BEV3DDetResult
 
 
-class BEVDet3DPredictor(BasicPredictor):
-    """BEVDet3DPredictor that inherits from BasicPredictor."""
+class BEVDet3DPredictor(BasePredictor):
+    """BEVDet3DPredictor that inherits from BasePredictor."""
 
     entities = MODELS
 
@@ -86,12 +80,11 @@ class BEVDet3DPredictor(BasicPredictor):
         Returns:
             tuple: A tuple containing the preprocessors and inference engine.
         """
-        if (
-            lazy_paddle.is_compiled_with_cuda()
-            and not lazy_paddle.is_compiled_with_rocm()
-        ):
-            from ....ops.voxelize import hard_voxelize
-            from ....ops.iou3d_nms import nms_gpu
+        import paddle
+
+        if paddle.is_compiled_with_cuda() and not paddle.is_compiled_with_rocm():
+            from ....ops.iou3d_nms import nms_gpu  # noqa: F401
+            from ....ops.voxelize import hard_voxelize  # noqa: F401
         else:
             logging.error("3D BEVFusion custom ops only support GPU platform!")
 
@@ -105,11 +98,7 @@ class BEVDet3DPredictor(BasicPredictor):
                 pre_tfs[name] = op
         pre_tfs["GetInferInput"] = GetInferInput()
 
-        infer = StaticInfer(
-            model_dir=self.model_dir,
-            model_prefix=self.MODEL_FILE_PREFIX,
-            option=self.pp_option,
-        )
+        infer = self.create_static_infer()
 
         return pre_tfs, infer
 

paddlex/inference/models/{3d_bev_detection → m_3d_bev_detection}/processors.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -14,25 +14,18 @@
 
 
 import numbers
-import cv2
+
 import numpy as np
-from typing import Generic, List, Optional
-import lazy_paddle as paddle
 
-from ...utils.io import ImageReader
-from ....utils import logging
+from ....utils.deps import class_requires_deps, is_dep_available
 from ...common.reader.det_3d_reader import Sample
+from ...utils.benchmark import benchmark
 
-
-cv2_interp_codes = {
-    "nearest": cv2.INTER_NEAREST,
-    "bilinear": cv2.INTER_LINEAR,
-    "bicubic": cv2.INTER_CUBIC,
-    "area": cv2.INTER_AREA,
-    "lanczos": cv2.INTER_LANCZOS4,
-}
+if is_dep_available("opencv-contrib-python"):
+    import cv2
 
 
+@benchmark.timeit
 class LoadPointsFromFile:
     """Load points from a file and process them according to specified parameters."""
 
@@ -47,6 +40,7 @@ class LoadPointsFromFile:
             shift_height (bool): Whether to shift height values.
             use_color (bool): Whether to include color attributes in the loaded points.
         """
+
         self.shift_height = shift_height
         self.use_color = use_color
         if isinstance(use_dim, int):
@@ -112,6 +106,7 @@ class LoadPointsFromFile:
         return results
 
 
+@benchmark.timeit
 class LoadPointsFromMultiSweeps(object):
     """Load points from multiple sweeps.This is usually used for nuScenes dataset to utilize previous sweeps."""
 
@@ -275,6 +270,8 @@ class LoadPointsFromMultiSweeps(object):
         return results
 
 
+@benchmark.timeit
+@class_requires_deps("opencv-contrib-python")
 class LoadMultiViewImageFromFiles:
     """Load multi-view images from files."""
 
@@ -341,6 +338,8 @@ class LoadMultiViewImageFromFiles:
         return sample
 
 
+@benchmark.timeit
+@class_requires_deps("opencv-contrib-python")
 class ResizeImage:
     """Resize images & bbox & mask."""
 
@@ -594,6 +593,13 @@ class ResizeImage:
         Returns:
             numpy.ndarray or tuple: The resized image. If return_scale is True, returns a tuple containing the resized image and the scaling factors (w_scale, h_scale).
         """
+        cv2_interp_codes = {
+            "nearest": cv2.INTER_NEAREST,
+            "bilinear": cv2.INTER_LINEAR,
+            "bicubic": cv2.INTER_CUBIC,
+            "area": cv2.INTER_AREA,
+            "lanczos": cv2.INTER_LANCZOS4,
+        }
         h, w = img.shape[:2]
         if backend not in ["cv2", "pillow"]:
             raise ValueError(
@@ -671,6 +677,8 @@ class ResizeImage:
         return results
 
 
+@benchmark.timeit
+@class_requires_deps("opencv-contrib-python")
 class NormalizeImage:
     """Normalize the image."""
 
@@ -729,6 +737,8 @@ class NormalizeImage:
         return results
 
 
+@benchmark.timeit
+@class_requires_deps("opencv-contrib-python")
 class PadImage(object):
     """Pad the image & mask."""
 
@@ -859,6 +869,7 @@ class PadImage(object):
         return results
 
 
+@benchmark.timeit
 class SampleFilterByKey:
     """Collect data from the loader relevant to the specific task."""
 
@@ -914,6 +925,7 @@ class SampleFilterByKey:
         return filtered_sample
 
 
+@benchmark.timeit
 class GetInferInput:
     """Collect infer input data from transformed sample"""
 

paddlex/inference/models/{3d_bev_detection → m_3d_bev_detection}/result.py

@@ -1,4 +1,4 @@
-# copyright (c) 2024 PaddlePaddle Authors. All Rights Reserve.
+# Copyright (c) 2024 PaddlePaddle Authors. All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -12,11 +12,14 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from ...common.result import BaseResult, StrMixin, JsonMixin
-import numpy as np
 import os
+
+import numpy as np
+
+from ...common.result import BaseResult
 from .visualizer_3d import Visualizer3D
 
+
 class BEV3DDetResult(BaseResult):
     """Base class for computer vision results."""
 
@@ -35,31 +38,31 @@ class BEV3DDetResult(BaseResult):
 
     def visualize(self, save_path: str, show: bool) -> None:
         # input point cloud
-        assert 'input_path' in self.keys(), 'input_path is not found in the data'
-        points = np.fromfile(self['input_path'], dtype=np.float32)
+        assert "input_path" in self.keys(), "input_path is not found in the data"
+        points = np.fromfile(self["input_path"], dtype=np.float32)
         points = points.reshape(-1, 5)
         points = points[:, :4]
 
         # detection result
         result = dict()
-        assert 'boxes_3d' in self.keys(), 'boxes_3d is not found in the data'
+        assert "boxes_3d" in self.keys(), "boxes_3d is not found in the data"
         result["bbox3d"] = self["boxes_3d"]
-        assert 'scores_3d' in self.keys(), 'scores_3d is not found in the data'
+        assert "scores_3d" in self.keys(), "scores_3d is not found in the data"
         result["scores"] = self["scores_3d"]
-        assert 'labels_3d' in self.keys(), 'labels_3d is not found in the data'
+        assert "labels_3d" in self.keys(), "labels_3d is not found in the data"
         result["labels"] = self["labels_3d"]
 
         if save_path is not None:
             # save result for local visualization
             if not os.path.exists(save_path):
                 os.makedirs(save_path)
-            np.save(os.path.join(save_path, 'results.npy'), result)
-            np.save(os.path.join(save_path, 'points.npy'), points)
-        
+            np.save(os.path.join(save_path, "results.npy"), result)
+            np.save(os.path.join(save_path, "points.npy"), points)
+
         if show:
             # visualize
             score_threshold = 0.25
             vis = Visualizer3D()
             vis.draw_results(points, result, score_threshold)
 
-        return
+        return