inference-models 0.18.3__py3-none-any.whl

inference_models/models/yolov8/yolov8_key_points_detection_onnx.py

@@ -0,0 +1,263 @@
+from threading import Lock
+from typing import List, Optional, Tuple, Union
+
+import numpy as np
+import torch
+
+from inference_models import Detections, KeyPoints, KeyPointsDetectionModel
+from inference_models.configuration import DEFAULT_DEVICE
+from inference_models.entities import ColorFormat
+from inference_models.errors import (
+    CorruptedModelPackageError,
+    EnvironmentConfigurationError,
+    MissingDependencyError,
+)
+from inference_models.models.common.model_packages import get_model_package_contents
+from inference_models.models.common.onnx import (
+    run_session_with_batch_size_limit,
+    set_execution_provider_defaults,
+)
+from inference_models.models.common.roboflow.model_packages import (
+    InferenceConfig,
+    PreProcessingMetadata,
+    ResizeMode,
+    parse_class_names_file,
+    parse_inference_config,
+    parse_key_points_metadata,
+)
+from inference_models.models.common.roboflow.post_processing import (
+    post_process_nms_fused_model_output,
+    rescale_key_points_detections,
+    run_nms_for_key_points_detection,
+)
+from inference_models.models.common.roboflow.pre_processing import (
+    pre_process_network_input,
+)
+from inference_models.utils.onnx_introspection import (
+    get_selected_onnx_execution_providers,
+)
+
+try:
+    import onnxruntime
+except ImportError as import_error:
+    raise MissingDependencyError(
+        message=f"Could not import YOLOv8 model with ONNX backend - this error means that some additional dependencies "
+        f"are not installed in the environment. If you run the `inference-models` library directly in your Python "
+        f"program, make sure the following extras of the package are installed: \n"
+        f"\t* `onnx-cpu` - when you wish to use library with CPU support only\n"
+        f"\t* `onnx-cu12` - for running on GPU with Cuda 12 installed\n"
+        f"\t* `onnx-cu118` - for running on GPU with Cuda 11.8 installed\n"
+        f"\t* `onnx-jp6-cu126` - for running on Jetson with Jetpack 6\n"
+        f"If you see this error using Roboflow infrastructure, make sure the service you use does support the model. "
+        f"You can also contact Roboflow to get support.",
+        help_url="https://todo",
+    ) from import_error
+
+
+class YOLOv8ForKeyPointsDetectionOnnx(
+    KeyPointsDetectionModel[torch.Tensor, PreProcessingMetadata, torch.Tensor]
+):
+
+    @classmethod
+    def from_pretrained(
+        cls,
+        model_name_or_path: str,
+        onnx_execution_providers: Optional[List[Union[str, tuple]]] = None,
+        default_onnx_trt_options: bool = True,
+        device: torch.device = DEFAULT_DEVICE,
+        **kwargs,
+    ) -> "YOLOv8ForKeyPointsDetectionOnnx":
+        if onnx_execution_providers is None:
+            onnx_execution_providers = get_selected_onnx_execution_providers()
+        if not onnx_execution_providers:
+            raise EnvironmentConfigurationError(
+                message=f"Could not initialize model - selected backend is ONNX which requires execution provider to "
+                f"be specified - explicitly in `from_pretrained(...)` method or via env variable "
+                f"`ONNXRUNTIME_EXECUTION_PROVIDERS`. If you run model locally - adjust your setup, otherwise "
+                f"contact the platform support.",
+                help_url="https://todo",
+            )
+        onnx_execution_providers = set_execution_provider_defaults(
+            providers=onnx_execution_providers,
+            model_package_path=model_name_or_path,
+            device=device,
+            default_onnx_trt_options=default_onnx_trt_options,
+        )
+        model_package_content = get_model_package_contents(
+            model_package_dir=model_name_or_path,
+            elements=[
+                "class_names.txt",
+                "inference_config.json",
+                "weights.onnx",
+                "keypoints_metadata.json",
+            ],
+        )
+        class_names = parse_class_names_file(
+            class_names_path=model_package_content["class_names.txt"]
+        )
+        inference_config = parse_inference_config(
+            config_path=model_package_content["inference_config.json"],
+            allowed_resize_modes={
+                ResizeMode.STRETCH_TO,
+                ResizeMode.LETTERBOX,
+                ResizeMode.CENTER_CROP,
+                ResizeMode.LETTERBOX_REFLECT_EDGES,
+            },
+        )
+        if inference_config.post_processing.type != "nms":
+            raise CorruptedModelPackageError(
+                message="Expected NMS to be the post-processing",
+                help_url="https://todo",
+            )
+        parsed_key_points_metadata, skeletons = parse_key_points_metadata(
+            key_points_metadata_path=model_package_content["keypoints_metadata.json"]
+        )
+        session = onnxruntime.InferenceSession(
+            path_or_bytes=model_package_content["weights.onnx"],
+            providers=onnx_execution_providers,
+        )
+        input_batch_size = session.get_inputs()[0].shape[0]
+        if isinstance(input_batch_size, str):
+            input_batch_size = None
+        input_name = session.get_inputs()[0].name
+        return cls(
+            session=session,
+            input_name=input_name,
+            class_names=class_names,
+            inference_config=inference_config,
+            device=device,
+            input_batch_size=input_batch_size,
+            parsed_key_points_metadata=parsed_key_points_metadata,
+            skeletons=skeletons,
+        )
+
+    def __init__(
+        self,
+        session: onnxruntime.InferenceSession,
+        input_name: str,
+        inference_config: InferenceConfig,
+        class_names: List[str],
+        device: torch.device,
+        input_batch_size: Optional[int],
+        parsed_key_points_metadata: List[List[str]],
+        skeletons: List[List[Tuple[int, int]]],
+    ):
+        self._session = session
+        self._input_name = input_name
+        self._inference_config = inference_config
+        self._class_names = class_names
+        self._skeletons = skeletons
+        self._device = device
+        self._input_batch_size = input_batch_size
+        self._session_thread_lock = Lock()
+        self._parsed_key_points_metadata = parsed_key_points_metadata
+        self._key_points_classes_for_instances = torch.tensor(
+            [len(e) for e in self._parsed_key_points_metadata], device=device
+        )
+        self._key_points_slots_in_prediction = max(
+            len(e) for e in parsed_key_points_metadata
+        )
+
+    @property
+    def class_names(self) -> List[str]:
+        return self._class_names
+
+    @property
+    def key_points_classes(self) -> List[List[str]]:
+        return self._parsed_key_points_metadata
+
+    @property
+    def skeletons(self) -> List[List[Tuple[int, int]]]:
+        return self._skeletons
+
+    def pre_process(
+        self,
+        images: Union[torch.Tensor, List[torch.Tensor], np.ndarray, List[np.ndarray]],
+        input_color_format: Optional[ColorFormat] = None,
+        image_size: Optional[Tuple[int, int]] = None,
+        **kwargs,
+    ) -> Tuple[torch.Tensor, List[PreProcessingMetadata]]:
+        return pre_process_network_input(
+            images=images,
+            image_pre_processing=self._inference_config.image_pre_processing,
+            network_input=self._inference_config.network_input,
+            target_device=self._device,
+            input_color_format=input_color_format,
+            image_size_wh=image_size,
+        )
+
+    def forward(self, pre_processed_images: torch.Tensor, **kwargs) -> torch.Tensor:
+        with self._session_thread_lock:
+            return run_session_with_batch_size_limit(
+                session=self._session,
+                inputs={self._input_name: pre_processed_images},
+                min_batch_size=self._input_batch_size,
+                max_batch_size=self._input_batch_size,
+            )[0]
+
+    def post_process(
+        self,
+        model_results: torch.Tensor,
+        pre_processing_meta: List[PreProcessingMetadata],
+        conf_thresh: float = 0.25,
+        iou_thresh: float = 0.45,
+        max_detections: int = 100,
+        class_agnostic: bool = False,
+        key_points_threshold: float = 0.3,
+        **kwargs,
+    ) -> Tuple[List[KeyPoints], Optional[List[Detections]]]:
+        if self._inference_config.post_processing.fused:
+            nms_results = post_process_nms_fused_model_output(
+                output=model_results, conf_thresh=conf_thresh
+            )
+        else:
+            nms_results = run_nms_for_key_points_detection(
+                output=model_results,
+                num_classes=len(self._class_names),
+                key_points_slots_in_prediction=self._key_points_slots_in_prediction,
+                conf_thresh=conf_thresh,
+                iou_thresh=iou_thresh,
+                max_detections=max_detections,
+                class_agnostic=class_agnostic,
+            )
+        rescaled_results = rescale_key_points_detections(
+            detections=nms_results,
+            images_metadata=pre_processing_meta,
+            num_classes=len(self._class_names),
+            key_points_slots_in_prediction=self._key_points_slots_in_prediction,
+        )
+        detections, all_key_points = [], []
+        for result in rescaled_results:
+            class_id = result[:, 5].int()
+            detections.append(
+                Detections(
+                    xyxy=result[:, :4].round().int(),
+                    class_id=class_id,
+                    confidence=result[:, 4],
+                )
+            )
+            key_points_reshaped = result[:, 6:].view(
+                result.shape[0], self._key_points_slots_in_prediction, 3
+            )
+            xy = key_points_reshaped[:, :, :2]
+            confidence = key_points_reshaped[:, :, 2]
+            key_points_classes_for_instance_class = (
+                (self._key_points_classes_for_instances[class_id])
+                .unsqueeze(1)
+                .to(device=result.device)
+            )
+            instances_class_mask = (
+                torch.arange(self._key_points_slots_in_prediction, device=result.device)
+                .unsqueeze(0)
+                .repeat(result.shape[0], 1)
+                < key_points_classes_for_instance_class
+            )
+
+            confidence_mask = confidence < key_points_threshold
+            mask = instances_class_mask & confidence_mask
+            xy[mask] = 0.0
+            confidence[mask] = 0.0
+            all_key_points.append(
+                KeyPoints(xy=xy.round().int(), class_id=class_id, confidence=confidence)
+            )
+        return all_key_points, detections

inference_models/models/yolov8/yolov8_key_points_detection_torch_script.py

@@ -0,0 +1,218 @@
+from typing import List, Optional, Tuple, Union
+
+import numpy as np
+import torch
+import torchvision  # DO NOT REMOVE, THIS IMPORT ENABLES NMS OPERATION
+
+from inference_models import Detections, KeyPoints, KeyPointsDetectionModel
+from inference_models.configuration import DEFAULT_DEVICE
+from inference_models.entities import ColorFormat
+from inference_models.errors import CorruptedModelPackageError
+from inference_models.models.common.model_packages import get_model_package_contents
+from inference_models.models.common.roboflow.model_packages import (
+    InferenceConfig,
+    PreProcessingMetadata,
+    ResizeMode,
+    parse_class_names_file,
+    parse_inference_config,
+    parse_key_points_metadata,
+)
+from inference_models.models.common.roboflow.post_processing import (
+    post_process_nms_fused_model_output,
+    rescale_key_points_detections,
+    run_nms_for_key_points_detection,
+)
+from inference_models.models.common.roboflow.pre_processing import (
+    pre_process_network_input,
+)
+from inference_models.models.common.torch import generate_batch_chunks
+
+
+class YOLOv8ForKeyPointsDetectionTorchScript(
+    KeyPointsDetectionModel[torch.Tensor, PreProcessingMetadata, torch.Tensor]
+):
+
+    @classmethod
+    def from_pretrained(
+        cls,
+        model_name_or_path: str,
+        device: torch.device = DEFAULT_DEVICE,
+        **kwargs,
+    ) -> "YOLOv8ForKeyPointsDetectionTorchScript":
+        model_package_content = get_model_package_contents(
+            model_package_dir=model_name_or_path,
+            elements=[
+                "class_names.txt",
+                "inference_config.json",
+                "weights.torchscript",
+                "keypoints_metadata.json",
+            ],
+        )
+        class_names = parse_class_names_file(
+            class_names_path=model_package_content["class_names.txt"]
+        )
+        inference_config = parse_inference_config(
+            config_path=model_package_content["inference_config.json"],
+            allowed_resize_modes={
+                ResizeMode.STRETCH_TO,
+                ResizeMode.LETTERBOX,
+                ResizeMode.CENTER_CROP,
+                ResizeMode.LETTERBOX_REFLECT_EDGES,
+            },
+        )
+        if inference_config.post_processing.type != "nms":
+            raise CorruptedModelPackageError(
+                message="Expected NMS to be the post-processing",
+                help_url="https://todo",
+            )
+        if inference_config.forward_pass.static_batch_size is None:
+            raise CorruptedModelPackageError(
+                message="Expected static batch size to be registered in the inference configuration.",
+                help_url="https://todo",
+            )
+        parsed_key_points_metadata, skeletons = parse_key_points_metadata(
+            key_points_metadata_path=model_package_content["keypoints_metadata.json"]
+        )
+        model = torch.jit.load(
+            model_package_content["weights.torchscript"], map_location=device
+        ).eval()
+        return cls(
+            model=model,
+            class_names=class_names,
+            inference_config=inference_config,
+            device=device,
+            parsed_key_points_metadata=parsed_key_points_metadata,
+            skeletons=skeletons,
+        )
+
+    def __init__(
+        self,
+        model: torch.nn.Module,
+        inference_config: InferenceConfig,
+        class_names: List[str],
+        device: torch.device,
+        parsed_key_points_metadata: List[List[str]],
+        skeletons: List[List[Tuple[int, int]]],
+    ):
+        self._model = model
+        self._inference_config = inference_config
+        self._class_names = class_names
+        self._skeletons = skeletons
+        self._device = device
+        self._parsed_key_points_metadata = parsed_key_points_metadata
+        self._key_points_classes_for_instances = torch.tensor(
+            [len(e) for e in self._parsed_key_points_metadata], device=device
+        )
+        self._key_points_slots_in_prediction = max(
+            len(e) for e in parsed_key_points_metadata
+        )
+
+    @property
+    def class_names(self) -> List[str]:
+        return self._class_names
+
+    @property
+    def key_points_classes(self) -> List[List[str]]:
+        return self._parsed_key_points_metadata
+
+    @property
+    def skeletons(self) -> List[List[Tuple[int, int]]]:
+        return self._skeletons
+
+    def pre_process(
+        self,
+        images: Union[torch.Tensor, List[torch.Tensor], np.ndarray, List[np.ndarray]],
+        input_color_format: Optional[ColorFormat] = None,
+        **kwargs,
+    ) -> Tuple[torch.Tensor, List[PreProcessingMetadata]]:
+        return pre_process_network_input(
+            images=images,
+            image_pre_processing=self._inference_config.image_pre_processing,
+            network_input=self._inference_config.network_input,
+            target_device=self._device,
+            input_color_format=input_color_format,
+        )
+
+    def forward(self, pre_processed_images: torch.Tensor, **kwargs) -> torch.Tensor:
+        with torch.inference_mode():
+            if (
+                pre_processed_images.shape[0]
+                == self._inference_config.forward_pass.static_batch_size
+            ):
+                return self._model(pre_processed_images).to(self._device)
+            results = []
+            for input_tensor, padding_size in generate_batch_chunks(
+                input_batch=pre_processed_images,
+                chunk_size=self._inference_config.forward_pass.static_batch_size,
+            ):
+                result_for_chunk = self._model(input_tensor)
+                if padding_size > 0:
+                    result_for_chunk = result_for_chunk[:-padding_size]
+                results.append(result_for_chunk)
+            return torch.cat(results, dim=0).to(self._device)
+
+    def post_process(
+        self,
+        model_results: torch.Tensor,
+        pre_processing_meta: List[PreProcessingMetadata],
+        conf_thresh: float = 0.25,
+        iou_thresh: float = 0.45,
+        max_detections: int = 100,
+        class_agnostic: bool = False,
+        key_points_threshold: float = 0.3,
+        **kwargs,
+    ) -> Tuple[List[KeyPoints], Optional[List[Detections]]]:
+        if self._inference_config.post_processing.fused:
+            nms_results = post_process_nms_fused_model_output(
+                output=model_results, conf_thresh=conf_thresh
+            )
+        else:
+            nms_results = run_nms_for_key_points_detection(
+                output=model_results,
+                num_classes=len(self._class_names),
+                key_points_slots_in_prediction=self._key_points_slots_in_prediction,
+                conf_thresh=conf_thresh,
+                iou_thresh=iou_thresh,
+                max_detections=max_detections,
+                class_agnostic=class_agnostic,
+            )
+        rescaled_results = rescale_key_points_detections(
+            detections=nms_results,
+            images_metadata=pre_processing_meta,
+            num_classes=len(self._class_names),
+            key_points_slots_in_prediction=self._key_points_slots_in_prediction,
+        )
+        detections, all_key_points = [], []
+        for result in rescaled_results:
+            class_id = result[:, 5].int()
+            detections.append(
+                Detections(
+                    xyxy=result[:, :4].round().int(),
+                    class_id=class_id,
+                    confidence=result[:, 4],
+                )
+            )
+            key_points_reshaped = result[:, 6:].view(
+                result.shape[0], self._key_points_slots_in_prediction, 3
+            )
+            xy = key_points_reshaped[:, :, :2]
+            confidence = key_points_reshaped[:, :, 2]
+            key_points_classes_for_instance_class = (
+                (self._key_points_classes_for_instances[class_id])
+                .unsqueeze(1)
+                .to(device=result.device)
+            )
+            instances_class_mask = (
+                torch.arange(self._key_points_slots_in_prediction, device=result.device)
+                .unsqueeze(0)
+                .repeat(result.shape[0], 1)
+                < key_points_classes_for_instance_class
+            )
+            confidence_mask = confidence < key_points_threshold
+            mask = instances_class_mask & confidence_mask
+            xy[mask] = 0.0
+            confidence[mask] = 0.0
+            all_key_points.append(
+                KeyPoints(xy=xy.round().int(), class_id=class_id, confidence=confidence)
+            )
+        return all_key_points, detections