first commit

tests/models/pp_ocrv5_server_rec/test_image_processing_pp_ocrv5_server_rec.py

@@ -0,0 +1,140 @@
+# Copyright 2026 The HuggingFace Inc. team. 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.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import math
+import unittest
+
+import numpy as np
+
+from transformers import is_vision_available
+from transformers.testing_utils import require_torch, require_vision
+
+from ...test_image_processing_common import ImageProcessingTestMixin, prepare_image_inputs
+
+
+if is_vision_available():
+    from PIL import Image
+
+
+class PPOCRV5ServerRecImageProcessingTester:
+    def __init__(
+        self,
+        parent,
+        batch_size=7,
+        num_channels=3,
+        image_size=18,
+        min_resolution=10,
+        max_resolution=400,
+        do_resize=True,
+        size=None,
+        do_rescale=True,
+        do_pad=True,
+        rescale_factor=1 / 255,
+        do_normalize=True,
+        max_image_width=3200,
+        image_mean=[0.5, 0.5, 0.5],
+        image_std=[0.5, 0.5, 0.5],
+    ):
+        size = size if size is not None else {"height": 48, "width": 320}
+        self.parent = parent
+        self.batch_size = batch_size
+        self.num_channels = num_channels
+        self.image_size = image_size
+        self.min_resolution = min_resolution
+        self.max_resolution = max_resolution
+        self.do_resize = do_resize
+        self.size = size
+        self.do_pad = do_pad
+        self.do_rescale = do_rescale
+        self.rescale_factor = rescale_factor
+        self.do_normalize = do_normalize
+        self.image_mean = image_mean
+        self.image_std = image_std
+        self.max_image_width = max_image_width
+
+    def prepare_image_processor_dict(self):
+        return {
+            "image_mean": self.image_mean,
+            "image_std": self.image_std,
+            "do_normalize": self.do_normalize,
+            "do_resize": self.do_resize,
+            "size": self.size,
+            "keep_aspect_ratio": False,
+            "do_pad": False,
+        }
+
+    def prepare_image_inputs(self, equal_resolution=False, numpify=False, torchify=False):
+        return prepare_image_inputs(
+            batch_size=self.batch_size,
+            num_channels=self.num_channels,
+            min_resolution=self.min_resolution,
+            max_resolution=self.max_resolution,
+            equal_resolution=equal_resolution,
+            numpify=numpify,
+            torchify=torchify,
+        )
+
+    def get_expected_value(self, images):
+        shape_list = []
+        for image in images:
+            if isinstance(image, Image.Image):
+                width, height = image.size
+            elif isinstance(image, np.ndarray):
+                height, width = image.shape[0], image.shape[1]
+            else:
+                height, width = image.shape[1], image.shape[2]
+            shape_list.append((height, width))
+
+        max_width = -1
+        max_height = -1
+        for height, width in shape_list:
+            # We need the width and height of the widest image in the batch
+            if width > max_width:
+                max_width = width
+                max_height = height
+
+        default_height, default_width = self.size["height"], self.size["width"]
+        ratio = max(max_width / max_height, default_width / default_height)
+
+        target_width = int(default_height * ratio)
+        target_height = default_height
+
+        if target_width > self.max_image_width:
+            target_width = self.max_image_width
+        else:
+            ratio = max_width / float(max_height)
+            if target_width >= math.ceil(default_height * ratio):
+                target_width = int(math.ceil(default_height * ratio))
+
+        return target_height, target_width
+
+    def expected_output_image_shape(self, images):
+        height, width = self.get_expected_value(images)
+        return self.num_channels, height, width
+
+
+@require_torch
+@require_vision
+class PPOCRV5ServerRecImageProcessingTest(ImageProcessingTestMixin, unittest.TestCase):
+    def setUp(self):
+        super().setUp()
+        self.image_processor_tester = PPOCRV5ServerRecImageProcessingTester(self)
+
+    @property
+    def image_processor_dict(self):
+        return self.image_processor_tester.prepare_image_processor_dict()
+
+    @unittest.skip(reason="PPOCRV5ServerRecImageProcessor does not support 4 channel images yet")
+    def test_call_numpy_4_channels():
+        pass

tests/models/pp_ocrv5_server_rec/test_modeling_pp_ocrv5_server_rec.py

@@ -0,0 +1,266 @@
+# coding = utf-8
+# Copyright 2026 The HuggingFace Inc. team. 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.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""Testing suite for the PPOCRV5ServerRec model."""
+
+import inspect
+import unittest
+
+from parameterized import parameterized
+
+from transformers import (
+    AutoImageProcessor,
+    AutoModelForTextRecognition,
+    PPOCRV5ServerRecConfig,
+    PPOCRV5ServerRecForTextRecognition,
+    is_torch_available,
+    is_vision_available,
+)
+from transformers.image_utils import load_image
+from transformers.testing_utils import (
+    require_torch,
+    require_torch_accelerator,
+    require_vision,
+    slow,
+    torch_device,
+)
+
+from ...test_configuration_common import ConfigTester
+from ...test_modeling_common import ModelTesterMixin, floats_tensor
+from ...test_pipeline_mixin import PipelineTesterMixin
+from ...test_processing_common import url_to_local_path
+
+
+if is_torch_available():
+    import torch
+
+
+class PPOCRV5ServerRecModelTester:
+    def __init__(
+        self,
+        parent,
+        batch_size=3,
+        image_size=[48, 320],
+        num_channels=3,
+        is_training=False,
+        hidden_act="silu",
+        hidden_size=10,
+        mlp_ratio=2.0,
+        depth=2,
+        head_out_channels=18385,
+        conv_kernel_size=[1, 3],
+        qkv_bias=True,
+        num_attention_heads=2,
+        attention_dropout=0.0,
+    ):
+        self.parent = parent
+        self.batch_size = batch_size
+        self.num_channels = num_channels
+        self.image_size = image_size
+        self.is_training = is_training
+        self.hidden_act = hidden_act
+        self.hidden_size = hidden_size
+        self.mlp_ratio = mlp_ratio
+        self.depth = depth
+        self.head_out_channels = head_out_channels
+        self.conv_kernel_size = conv_kernel_size
+        self.qkv_bias = qkv_bias
+        self.num_attention_heads = num_attention_heads
+        self.attention_dropout = attention_dropout
+
+    def prepare_config_and_inputs_for_common(self):
+        config, pixel_values = self.prepare_config_and_inputs()
+        inputs_dict = {"pixel_values": pixel_values}
+        return config, inputs_dict
+
+    def prepare_config_and_inputs(self):
+        pixel_values = floats_tensor([self.batch_size, self.num_channels, self.image_size[0], self.image_size[1]])
+        config = self.get_config()
+
+        return config, pixel_values
+
+    def get_config(self) -> PPOCRV5ServerRecConfig:
+        backbone_config = {
+            "model_type": "hgnet_v2",
+            "arch": "L",
+            "return_idx": [0, 1, 2, 3],
+            "hidden_sizes": [16, 16, 16, 16],
+            "stem_channels": [3, 16, 16],
+            "stage_in_channels": [16, 16, 16, 16],
+            "stage_mid_channels": [16, 16, 16, 16],
+            "stage_out_channels": [16, 16, 16, 16],
+            "freeze_stem_only": True,
+            "freeze_at": 0,
+            "freeze_norm": True,
+            "lr_mult_list": [1.0, 1.0, 1.0, 1.0, 1.0],
+            "out_features": ["stage1", "stage2", "stage3", "stage4"],
+            "stage_downsample": [True, True, True, True],
+            "stem_strides": [2, 1, 1, 1, 1],
+            "stage_downsample_strides": [[2, 1], [1, 2], [2, 1], [2, 1]],
+        }
+
+        config = PPOCRV5ServerRecConfig(
+            backbone_config=backbone_config,
+            hidden_act=self.hidden_act,
+            hidden_size=self.hidden_size,
+            mlp_ratio=self.mlp_ratio,
+            depth=self.depth,
+            head_out_channels=self.head_out_channels,
+            conv_kernel_size=self.conv_kernel_size,
+            qkv_bias=self.qkv_bias,
+            num_attention_heads=self.num_attention_heads,
+            attention_dropout=self.attention_dropout,
+        )
+
+        return config
+
+
+@require_torch
+class PPOCRV5ServerRecModelTest(ModelTesterMixin, PipelineTesterMixin, unittest.TestCase):
+    all_model_classes = (PPOCRV5ServerRecForTextRecognition,) if is_torch_available() else ()
+    pipeline_model_mapping = (
+        {"image-feature-extraction": PPOCRV5ServerRecForTextRecognition} if is_torch_available() else {}
+    )
+
+    has_attentions = False
+    test_resize_embeddings = False
+    model_split_percents = [0.5, 0.8]
+
+    def setUp(self):
+        self.model_tester = PPOCRV5ServerRecModelTester(self)
+        self.config_tester = ConfigTester(
+            self,
+            config_class=PPOCRV5ServerRecConfig,
+            has_text_modality=False,
+            common_properties=[],
+        )
+
+    def test_config(self):
+        self.config_tester.run_common_tests()
+
+    @unittest.skip("PPOCRV5ServerRec does not has no attribute `hf_device_map`")
+    def test_model_parallelism(self):
+        pass
+
+    @unittest.skip("PPOCRV5ServerRec does not has no function `get_input_embeddings`")
+    def test_model_get_set_embeddings(self):
+        pass
+
+    @unittest.skip(reason="PPOCRV5ServerRec does not support attention")
+    def test_retain_grad_hidden_states_attentions(self):
+        pass
+
+    def test_forward_signature(self):
+        config, _ = self.model_tester.prepare_config_and_inputs_for_common()
+
+        for model_class in self.all_model_classes:
+            model = model_class(config)
+            signature = inspect.signature(model.forward)
+            arg_names = [*signature.parameters.keys()]
+            expected_arg_names = ["pixel_values"]
+            self.assertListEqual(arg_names[:1], expected_arg_names)
+
+    @parameterized.expand(["float32", "float16", "bfloat16"])
+    @require_torch_accelerator
+    @slow
+    def test_inference_with_different_dtypes(self, dtype_str):
+        dtype = {
+            "float32": torch.float32,
+            "float16": torch.float16,
+            "bfloat16": torch.bfloat16,
+        }[dtype_str]
+
+        config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
+
+        for model_class in self.all_model_classes:
+            model = model_class(config)
+            model.to(torch_device).to(dtype)
+            model.eval()
+            for key, tensor in inputs_dict.items():
+                if tensor.dtype == torch.float32:
+                    inputs_dict[key] = tensor.to(dtype)
+            with torch.no_grad():
+                _ = model(**self._prepare_for_class(inputs_dict, model_class))
+
+    def test_hidden_states_output(self):
+        # PP-OCRV5 uses HGNetV2 backbone: hidden_states = (embedding, stage1, ..., stageN) = num_stages + 1
+        # and head_hidden_states = config.depth + 1
+        config = self.model_tester.get_config()
+        num_expected_hidden_states = len(config.backbone_config.depths) + 1
+
+        def check_hidden_states_output(inputs_dict, config, model_class):
+            model = model_class(config)
+            model.to(torch_device)
+            model.eval()
+
+            with torch.no_grad():
+                outputs = model(**self._prepare_for_class(inputs_dict, model_class))
+
+            hidden_states = outputs.hidden_states
+            self.assertIsNotNone(hidden_states)
+            self.assertEqual(len(hidden_states), num_expected_hidden_states)
+
+            # First hidden state (embedding output) is 2x downsampled
+            self.assertListEqual(
+                list(hidden_states[0].shape[-2:]),
+                [self.model_tester.image_size[0] // 2, self.model_tester.image_size[1] // 2],
+            )
+
+            head_hidden_states = outputs.head_hidden_states
+            self.assertIsNotNone(head_hidden_states)
+            self.assertEqual(len(head_hidden_states), self.model_tester.depth + 1)
+
+            self.assertListEqual(
+                list(head_hidden_states[0].shape[-2:]),
+                [self.model_tester.hidden_size * self.model_tester.mlp_ratio * 2, self.model_tester.hidden_size],
+            )
+
+        config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common()
+        for model_class in self.all_model_classes:
+            inputs_dict["output_hidden_states"] = True
+            check_hidden_states_output(inputs_dict.copy(), config, model_class)
+
+            # Check that output_hidden_states also works via config (including backbone subconfig)
+            del inputs_dict["output_hidden_states"]
+            config.output_hidden_states = True
+            if config.backbone_config is not None:
+                config.backbone_config.output_hidden_states = True
+            check_hidden_states_output(inputs_dict.copy(), config, model_class)
+
+
+@require_torch
+@require_vision
+@slow
+class PPOCRV5ServerRecModelIntegrationTest(unittest.TestCase):
+    def setUp(self):
+        model_path = "PaddlePaddle/PP-OCRv5_server_rec_safetensors"
+        self.model = AutoModelForTextRecognition.from_pretrained(model_path).to(torch_device)
+        self.image_processor = (
+            AutoImageProcessor.from_pretrained(model_path, return_tensors="pt") if is_vision_available() else None
+        )
+        img_url = url_to_local_path(
+            "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_ocr_rec_001.png"
+        )
+        self.image = load_image(img_url)
+
+    def test_inference_text_recognition_head(self):
+        inputs = self.image_processor(images=self.image, return_tensors="pt").to(torch_device)
+        outputs = self.model(**inputs)
+
+        results = self.image_processor.post_process_text_recognition(outputs)
+        expected_text = "绿洲仕格维花园公寓"
+        expected_score = 0.9837473630905151
+
+        self.assertEqual(results[0]["text"], expected_text)
+        torch.testing.assert_close(results[0]["score"], expected_score, rtol=2e-2, atol=2e-2)