opencv-contrib-python 4.12.0.88__cp37-abi3-manylinux2014_aarch64.manylinux_2_17_aarch64.whl

cv2/dnn/__init__.pyi

@@ -0,0 +1,536 @@
+__all__: list[str] = []
+
+import cv2
+import cv2.typing
+import numpy
+import sys
+import typing as _typing
+if sys.version_info >= (3, 8):
+    from typing import Protocol
+else:
+    from typing_extensions import Protocol
+
+
+# Enumerations
+DNN_BACKEND_DEFAULT: int
+DNN_BACKEND_HALIDE: int
+DNN_BACKEND_INFERENCE_ENGINE: int
+DNN_BACKEND_OPENCV: int
+DNN_BACKEND_VKCOM: int
+DNN_BACKEND_CUDA: int
+DNN_BACKEND_WEBNN: int
+DNN_BACKEND_TIMVX: int
+DNN_BACKEND_CANN: int
+Backend = int
+"""One of [DNN_BACKEND_DEFAULT, DNN_BACKEND_HALIDE, DNN_BACKEND_INFERENCE_ENGINE, DNN_BACKEND_OPENCV, DNN_BACKEND_VKCOM, DNN_BACKEND_CUDA, DNN_BACKEND_WEBNN, DNN_BACKEND_TIMVX, DNN_BACKEND_CANN]"""
+
+DNN_TARGET_CPU: int
+DNN_TARGET_OPENCL: int
+DNN_TARGET_OPENCL_FP16: int
+DNN_TARGET_MYRIAD: int
+DNN_TARGET_VULKAN: int
+DNN_TARGET_FPGA: int
+DNN_TARGET_CUDA: int
+DNN_TARGET_CUDA_FP16: int
+DNN_TARGET_HDDL: int
+DNN_TARGET_NPU: int
+DNN_TARGET_CPU_FP16: int
+Target = int
+"""One of [DNN_TARGET_CPU, DNN_TARGET_OPENCL, DNN_TARGET_OPENCL_FP16, DNN_TARGET_MYRIAD, DNN_TARGET_VULKAN, DNN_TARGET_FPGA, DNN_TARGET_CUDA, DNN_TARGET_CUDA_FP16, DNN_TARGET_HDDL, DNN_TARGET_NPU, DNN_TARGET_CPU_FP16]"""
+
+DNN_LAYOUT_UNKNOWN: int
+DNN_LAYOUT_ND: int
+DNN_LAYOUT_NCHW: int
+DNN_LAYOUT_NCDHW: int
+DNN_LAYOUT_NHWC: int
+DNN_LAYOUT_NDHWC: int
+DNN_LAYOUT_PLANAR: int
+DataLayout = int
+"""One of [DNN_LAYOUT_UNKNOWN, DNN_LAYOUT_ND, DNN_LAYOUT_NCHW, DNN_LAYOUT_NCDHW, DNN_LAYOUT_NHWC, DNN_LAYOUT_NDHWC, DNN_LAYOUT_PLANAR]"""
+
+DNN_PMODE_NULL: int
+DNN_PMODE_CROP_CENTER: int
+DNN_PMODE_LETTERBOX: int
+ImagePaddingMode = int
+"""One of [DNN_PMODE_NULL, DNN_PMODE_CROP_CENTER, DNN_PMODE_LETTERBOX]"""
+
+SoftNMSMethod_SOFTNMS_LINEAR: int
+SOFT_NMSMETHOD_SOFTNMS_LINEAR: int
+SoftNMSMethod_SOFTNMS_GAUSSIAN: int
+SOFT_NMSMETHOD_SOFTNMS_GAUSSIAN: int
+SoftNMSMethod = int
+"""One of [SoftNMSMethod_SOFTNMS_LINEAR, SOFT_NMSMETHOD_SOFTNMS_LINEAR, SoftNMSMethod_SOFTNMS_GAUSSIAN, SOFT_NMSMETHOD_SOFTNMS_GAUSSIAN]"""
+
+
+
+# Classes
+class DictValue:
+    # Functions
+    @_typing.overload
+    def __init__(self, i: int) -> None: ...
+    @_typing.overload
+    def __init__(self, p: float) -> None: ...
+    @_typing.overload
+    def __init__(self, s: str) -> None: ...
+
+    def isInt(self) -> bool: ...
+
+    def isString(self) -> bool: ...
+
+    def isReal(self) -> bool: ...
+
+    def getIntValue(self, idx: int = ...) -> int: ...
+
+    def getRealValue(self, idx: int = ...) -> float: ...
+
+    def getStringValue(self, idx: int = ...) -> str: ...
+
+
+class Layer(cv2.Algorithm):
+    blobs: _typing.Sequence[cv2.typing.MatLike]
+    @property
+    def name(self) -> str: ...
+    @property
+    def type(self) -> str: ...
+    @property
+    def preferableTarget(self) -> int: ...
+
+    # Functions
+    @_typing.overload
+    def finalize(self, inputs: _typing.Sequence[cv2.typing.MatLike], outputs: _typing.Sequence[cv2.typing.MatLike] | None = ...) -> _typing.Sequence[cv2.typing.MatLike]: ...
+    @_typing.overload
+    def finalize(self, inputs: _typing.Sequence[cv2.UMat], outputs: _typing.Sequence[cv2.UMat] | None = ...) -> _typing.Sequence[cv2.UMat]: ...
+
+    def run(self, inputs: _typing.Sequence[cv2.typing.MatLike], internals: _typing.Sequence[cv2.typing.MatLike], outputs: _typing.Sequence[cv2.typing.MatLike] | None = ...) -> tuple[_typing.Sequence[cv2.typing.MatLike], _typing.Sequence[cv2.typing.MatLike]]: ...
+
+    def outputNameToIndex(self, outputName: str) -> int: ...
+
+
+class Net:
+    # Functions
+    def __init__(self) -> None: ...
+
+    @classmethod
+    @_typing.overload
+    def readFromModelOptimizer(cls, xml: str, bin: str) -> Net: ...
+    @classmethod
+    @_typing.overload
+    def readFromModelOptimizer(cls, bufferModelConfig: numpy.ndarray[_typing.Any, numpy.dtype[numpy.uint8]], bufferWeights: numpy.ndarray[_typing.Any, numpy.dtype[numpy.uint8]]) -> Net: ...
+
+    def empty(self) -> bool: ...
+
+    def dump(self) -> str: ...
+
+    def dumpToFile(self, path: str) -> None: ...
+
+    def dumpToPbtxt(self, path: str) -> None: ...
+
+    def addLayer(self, name: str, type: str, dtype: int, params: cv2.typing.LayerParams) -> int: ...
+
+    def addLayerToPrev(self, name: str, type: str, dtype: int, params: cv2.typing.LayerParams) -> int: ...
+
+    def getLayerId(self, layer: str) -> int: ...
+
+    def getLayerNames(self) -> _typing.Sequence[str]: ...
+
+    @_typing.overload
+    def getLayer(self, layerId: int) -> Layer: ...
+    @_typing.overload
+    def getLayer(self, layerName: str) -> Layer: ...
+    @_typing.overload
+    def getLayer(self, layerId: cv2.typing.LayerId) -> Layer: ...
+
+    def connect(self, outPin: str, inpPin: str) -> None: ...
+
+    def registerOutput(self, outputName: str, layerId: int, outputPort: int) -> int: ...
+
+    def setInputsNames(self, inputBlobNames: _typing.Sequence[str]) -> None: ...
+
+    def setInputShape(self, inputName: str, shape: cv2.typing.MatShape) -> None: ...
+
+    @_typing.overload
+    def forward(self, outputName: str = ...) -> cv2.typing.MatLike: ...
+    @_typing.overload
+    def forward(self, outputBlobs: _typing.Sequence[cv2.typing.MatLike] | None = ..., outputName: str = ...) -> _typing.Sequence[cv2.typing.MatLike]: ...
+    @_typing.overload
+    def forward(self, outputBlobs: _typing.Sequence[cv2.UMat] | None = ..., outputName: str = ...) -> _typing.Sequence[cv2.UMat]: ...
+    @_typing.overload
+    def forward(self, outBlobNames: _typing.Sequence[str], outputBlobs: _typing.Sequence[cv2.typing.MatLike] | None = ...) -> _typing.Sequence[cv2.typing.MatLike]: ...
+    @_typing.overload
+    def forward(self, outBlobNames: _typing.Sequence[str], outputBlobs: _typing.Sequence[cv2.UMat] | None = ...) -> _typing.Sequence[cv2.UMat]: ...
+
+    def forwardAsync(self, outputName: str = ...) -> cv2.AsyncArray: ...
+
+    def forwardAndRetrieve(self, outBlobNames: _typing.Sequence[str]) -> _typing.Sequence[_typing.Sequence[cv2.typing.MatLike]]: ...
+
+    @_typing.overload
+    def quantize(self, calibData: _typing.Sequence[cv2.typing.MatLike], inputsDtype: int, outputsDtype: int, perChannel: bool = ...) -> Net: ...
+    @_typing.overload
+    def quantize(self, calibData: _typing.Sequence[cv2.UMat], inputsDtype: int, outputsDtype: int, perChannel: bool = ...) -> Net: ...
+
+    def getInputDetails(self) -> tuple[_typing.Sequence[float], _typing.Sequence[int]]: ...
+
+    def getOutputDetails(self) -> tuple[_typing.Sequence[float], _typing.Sequence[int]]: ...
+
+    def setHalideScheduler(self, scheduler: str) -> None: ...
+
+    def setPreferableBackend(self, backendId: int) -> None: ...
+
+    def setPreferableTarget(self, targetId: int) -> None: ...
+
+    @_typing.overload
+    def setInput(self, blob: cv2.typing.MatLike, name: str = ..., scalefactor: float = ..., mean: cv2.typing.Scalar = ...) -> None: ...
+    @_typing.overload
+    def setInput(self, blob: cv2.UMat, name: str = ..., scalefactor: float = ..., mean: cv2.typing.Scalar = ...) -> None: ...
+
+    @_typing.overload
+    def setParam(self, layer: int, numParam: int, blob: cv2.typing.MatLike) -> None: ...
+    @_typing.overload
+    def setParam(self, layerName: str, numParam: int, blob: cv2.typing.MatLike) -> None: ...
+
+    @_typing.overload
+    def getParam(self, layer: int, numParam: int = ...) -> cv2.typing.MatLike: ...
+    @_typing.overload
+    def getParam(self, layerName: str, numParam: int = ...) -> cv2.typing.MatLike: ...
+
+    def getUnconnectedOutLayers(self) -> _typing.Sequence[int]: ...
+
+    def getUnconnectedOutLayersNames(self) -> _typing.Sequence[str]: ...
+
+    @_typing.overload
+    def getLayersShapes(self, netInputShapes: _typing.Sequence[cv2.typing.MatShape]) -> tuple[_typing.Sequence[int], _typing.Sequence[_typing.Sequence[cv2.typing.MatShape]], _typing.Sequence[_typing.Sequence[cv2.typing.MatShape]]]: ...
+    @_typing.overload
+    def getLayersShapes(self, netInputShape: cv2.typing.MatShape) -> tuple[_typing.Sequence[int], _typing.Sequence[_typing.Sequence[cv2.typing.MatShape]], _typing.Sequence[_typing.Sequence[cv2.typing.MatShape]]]: ...
+
+    @_typing.overload
+    def getFLOPS(self, netInputShapes: _typing.Sequence[cv2.typing.MatShape]) -> int: ...
+    @_typing.overload
+    def getFLOPS(self, netInputShape: cv2.typing.MatShape) -> int: ...
+    @_typing.overload
+    def getFLOPS(self, layerId: int, netInputShapes: _typing.Sequence[cv2.typing.MatShape]) -> int: ...
+    @_typing.overload
+    def getFLOPS(self, layerId: int, netInputShape: cv2.typing.MatShape) -> int: ...
+
+    def getLayerTypes(self) -> _typing.Sequence[str]: ...
+
+    def getLayersCount(self, layerType: str) -> int: ...
+
+    @_typing.overload
+    def getMemoryConsumption(self, netInputShape: cv2.typing.MatShape) -> tuple[int, int]: ...
+    @_typing.overload
+    def getMemoryConsumption(self, layerId: int, netInputShapes: _typing.Sequence[cv2.typing.MatShape]) -> tuple[int, int]: ...
+    @_typing.overload
+    def getMemoryConsumption(self, layerId: int, netInputShape: cv2.typing.MatShape) -> tuple[int, int]: ...
+
+    def enableFusion(self, fusion: bool) -> None: ...
+
+    def enableWinograd(self, useWinograd: bool) -> None: ...
+
+    def getPerfProfile(self) -> tuple[int, _typing.Sequence[float]]: ...
+
+
+class Image2BlobParams:
+    scalefactor: cv2.typing.Scalar
+    size: cv2.typing.Size
+    mean: cv2.typing.Scalar
+    swapRB: bool
+    ddepth: int
+    datalayout: DataLayout
+    paddingmode: ImagePaddingMode
+    borderValue: cv2.typing.Scalar
+
+    # Functions
+    @_typing.overload
+    def __init__(self) -> None: ...
+    @_typing.overload
+    def __init__(self, scalefactor: cv2.typing.Scalar, size: cv2.typing.Size = ..., mean: cv2.typing.Scalar = ..., swapRB: bool = ..., ddepth: int = ..., datalayout: DataLayout = ..., mode: ImagePaddingMode = ..., borderValue: cv2.typing.Scalar = ...) -> None: ...
+
+    def blobRectToImageRect(self, rBlob: cv2.typing.Rect, size: cv2.typing.Size) -> cv2.typing.Rect: ...
+
+    def blobRectsToImageRects(self, rBlob: _typing.Sequence[cv2.typing.Rect], size: cv2.typing.Size) -> _typing.Sequence[cv2.typing.Rect]: ...
+
+
+class Model:
+    # Functions
+    @_typing.overload
+    def __init__(self, model: str, config: str = ...) -> None: ...
+    @_typing.overload
+    def __init__(self, network: Net) -> None: ...
+
+    @_typing.overload
+    def setInputSize(self, size: cv2.typing.Size) -> Model: ...
+    @_typing.overload
+    def setInputSize(self, width: int, height: int) -> Model: ...
+
+    def setInputMean(self, mean: cv2.typing.Scalar) -> Model: ...
+
+    def setInputScale(self, scale: cv2.typing.Scalar) -> Model: ...
+
+    def setInputCrop(self, crop: bool) -> Model: ...
+
+    def setInputSwapRB(self, swapRB: bool) -> Model: ...
+
+    def setOutputNames(self, outNames: _typing.Sequence[str]) -> Model: ...
+
+    def setInputParams(self, scale: float = ..., size: cv2.typing.Size = ..., mean: cv2.typing.Scalar = ..., swapRB: bool = ..., crop: bool = ...) -> None: ...
+
+    @_typing.overload
+    def predict(self, frame: cv2.typing.MatLike, outs: _typing.Sequence[cv2.typing.MatLike] | None = ...) -> _typing.Sequence[cv2.typing.MatLike]: ...
+    @_typing.overload
+    def predict(self, frame: cv2.UMat, outs: _typing.Sequence[cv2.UMat] | None = ...) -> _typing.Sequence[cv2.UMat]: ...
+
+    def setPreferableBackend(self, backendId: Backend) -> Model: ...
+
+    def setPreferableTarget(self, targetId: Target) -> Model: ...
+
+    def enableWinograd(self, useWinograd: bool) -> Model: ...
+
+
+class ClassificationModel(Model):
+    # Functions
+    @_typing.overload
+    def __init__(self, model: str, config: str = ...) -> None: ...
+    @_typing.overload
+    def __init__(self, network: Net) -> None: ...
+
+    def setEnableSoftmaxPostProcessing(self, enable: bool) -> ClassificationModel: ...
+
+    def getEnableSoftmaxPostProcessing(self) -> bool: ...
+
+    @_typing.overload
+    def classify(self, frame: cv2.typing.MatLike) -> tuple[int, float]: ...
+    @_typing.overload
+    def classify(self, frame: cv2.UMat) -> tuple[int, float]: ...
+
+
+class KeypointsModel(Model):
+    # Functions
+    @_typing.overload
+    def __init__(self, model: str, config: str = ...) -> None: ...
+    @_typing.overload
+    def __init__(self, network: Net) -> None: ...
+
+    @_typing.overload
+    def estimate(self, frame: cv2.typing.MatLike, thresh: float = ...) -> _typing.Sequence[cv2.typing.Point2f]: ...
+    @_typing.overload
+    def estimate(self, frame: cv2.UMat, thresh: float = ...) -> _typing.Sequence[cv2.typing.Point2f]: ...
+
+
+class SegmentationModel(Model):
+    # Functions
+    @_typing.overload
+    def __init__(self, model: str, config: str = ...) -> None: ...
+    @_typing.overload
+    def __init__(self, network: Net) -> None: ...
+
+    @_typing.overload
+    def segment(self, frame: cv2.typing.MatLike, mask: cv2.typing.MatLike | None = ...) -> cv2.typing.MatLike: ...
+    @_typing.overload
+    def segment(self, frame: cv2.UMat, mask: cv2.UMat | None = ...) -> cv2.UMat: ...
+
+
+class DetectionModel(Model):
+    # Functions
+    @_typing.overload
+    def __init__(self, model: str, config: str = ...) -> None: ...
+    @_typing.overload
+    def __init__(self, network: Net) -> None: ...
+
+    def setNmsAcrossClasses(self, value: bool) -> DetectionModel: ...
+
+    def getNmsAcrossClasses(self) -> bool: ...
+
+    @_typing.overload
+    def detect(self, frame: cv2.typing.MatLike, confThreshold: float = ..., nmsThreshold: float = ...) -> tuple[_typing.Sequence[int], _typing.Sequence[float], _typing.Sequence[cv2.typing.Rect]]: ...
+    @_typing.overload
+    def detect(self, frame: cv2.UMat, confThreshold: float = ..., nmsThreshold: float = ...) -> tuple[_typing.Sequence[int], _typing.Sequence[float], _typing.Sequence[cv2.typing.Rect]]: ...
+
+
+class TextRecognitionModel(Model):
+    # Functions
+    @_typing.overload
+    def __init__(self, network: Net) -> None: ...
+    @_typing.overload
+    def __init__(self, model: str, config: str = ...) -> None: ...
+
+    def setDecodeType(self, decodeType: str) -> TextRecognitionModel: ...
+
+    def getDecodeType(self) -> str: ...
+
+    def setDecodeOptsCTCPrefixBeamSearch(self, beamSize: int, vocPruneSize: int = ...) -> TextRecognitionModel: ...
+
+    def setVocabulary(self, vocabulary: _typing.Sequence[str]) -> TextRecognitionModel: ...
+
+    def getVocabulary(self) -> _typing.Sequence[str]: ...
+
+    @_typing.overload
+    def recognize(self, frame: cv2.typing.MatLike) -> str: ...
+    @_typing.overload
+    def recognize(self, frame: cv2.UMat) -> str: ...
+    @_typing.overload
+    def recognize(self, frame: cv2.typing.MatLike, roiRects: _typing.Sequence[cv2.typing.MatLike]) -> _typing.Sequence[str]: ...
+    @_typing.overload
+    def recognize(self, frame: cv2.UMat, roiRects: _typing.Sequence[cv2.UMat]) -> _typing.Sequence[str]: ...
+
+
+class TextDetectionModel(Model):
+    # Functions
+    @_typing.overload
+    def detect(self, frame: cv2.typing.MatLike) -> tuple[_typing.Sequence[_typing.Sequence[cv2.typing.Point]], _typing.Sequence[float]]: ...
+    @_typing.overload
+    def detect(self, frame: cv2.UMat) -> tuple[_typing.Sequence[_typing.Sequence[cv2.typing.Point]], _typing.Sequence[float]]: ...
+    @_typing.overload
+    def detect(self, frame: cv2.typing.MatLike) -> _typing.Sequence[_typing.Sequence[cv2.typing.Point]]: ...
+    @_typing.overload
+    def detect(self, frame: cv2.UMat) -> _typing.Sequence[_typing.Sequence[cv2.typing.Point]]: ...
+
+    @_typing.overload
+    def detectTextRectangles(self, frame: cv2.typing.MatLike) -> tuple[_typing.Sequence[cv2.typing.RotatedRect], _typing.Sequence[float]]: ...
+    @_typing.overload
+    def detectTextRectangles(self, frame: cv2.UMat) -> tuple[_typing.Sequence[cv2.typing.RotatedRect], _typing.Sequence[float]]: ...
+    @_typing.overload
+    def detectTextRectangles(self, frame: cv2.typing.MatLike) -> _typing.Sequence[cv2.typing.RotatedRect]: ...
+    @_typing.overload
+    def detectTextRectangles(self, frame: cv2.UMat) -> _typing.Sequence[cv2.typing.RotatedRect]: ...
+
+
+class TextDetectionModel_EAST(TextDetectionModel):
+    # Functions
+    @_typing.overload
+    def __init__(self, network: Net) -> None: ...
+    @_typing.overload
+    def __init__(self, model: str, config: str = ...) -> None: ...
+
+    def setConfidenceThreshold(self, confThreshold: float) -> TextDetectionModel_EAST: ...
+
+    def getConfidenceThreshold(self) -> float: ...
+
+    def setNMSThreshold(self, nmsThreshold: float) -> TextDetectionModel_EAST: ...
+
+    def getNMSThreshold(self) -> float: ...
+
+
+class TextDetectionModel_DB(TextDetectionModel):
+    # Functions
+    @_typing.overload
+    def __init__(self, network: Net) -> None: ...
+    @_typing.overload
+    def __init__(self, model: str, config: str = ...) -> None: ...
+
+    def setBinaryThreshold(self, binaryThreshold: float) -> TextDetectionModel_DB: ...
+
+    def getBinaryThreshold(self) -> float: ...
+
+    def setPolygonThreshold(self, polygonThreshold: float) -> TextDetectionModel_DB: ...
+
+    def getPolygonThreshold(self) -> float: ...
+
+    def setUnclipRatio(self, unclipRatio: float) -> TextDetectionModel_DB: ...
+
+    def getUnclipRatio(self) -> float: ...
+
+    def setMaxCandidates(self, maxCandidates: int) -> TextDetectionModel_DB: ...
+
+    def getMaxCandidates(self) -> int: ...
+
+
+class LayerProtocol(Protocol):
+    # Functions
+    def __init__(self, params: dict[str, DictValue], blobs: _typing.Sequence[cv2.typing.MatLike]) -> None: ...
+
+    def getMemoryShapes(self, inputs: _typing.Sequence[_typing.Sequence[int]]) -> _typing.Sequence[_typing.Sequence[int]]: ...
+
+    def forward(self, inputs: _typing.Sequence[cv2.typing.MatLike]) -> _typing.Sequence[cv2.typing.MatLike]: ...
+
+
+
+# Functions
+def NMSBoxes(bboxes: _typing.Sequence[cv2.typing.Rect2d], scores: _typing.Sequence[float], score_threshold: float, nms_threshold: float, eta: float = ..., top_k: int = ...) -> _typing.Sequence[int]: ...
+
+def NMSBoxesBatched(bboxes: _typing.Sequence[cv2.typing.Rect2d], scores: _typing.Sequence[float], class_ids: _typing.Sequence[int], score_threshold: float, nms_threshold: float, eta: float = ..., top_k: int = ...) -> _typing.Sequence[int]: ...
+
+def NMSBoxesRotated(bboxes: _typing.Sequence[cv2.typing.RotatedRect], scores: _typing.Sequence[float], score_threshold: float, nms_threshold: float, eta: float = ..., top_k: int = ...) -> _typing.Sequence[int]: ...
+
+@_typing.overload
+def blobFromImage(image: cv2.typing.MatLike, scalefactor: float = ..., size: cv2.typing.Size = ..., mean: cv2.typing.Scalar = ..., swapRB: bool = ..., crop: bool = ..., ddepth: int = ...) -> cv2.typing.MatLike: ...
+@_typing.overload
+def blobFromImage(image: cv2.UMat, scalefactor: float = ..., size: cv2.typing.Size = ..., mean: cv2.typing.Scalar = ..., swapRB: bool = ..., crop: bool = ..., ddepth: int = ...) -> cv2.typing.MatLike: ...
+
+@_typing.overload
+def blobFromImageWithParams(image: cv2.typing.MatLike, param: Image2BlobParams = ...) -> cv2.typing.MatLike: ...
+@_typing.overload
+def blobFromImageWithParams(image: cv2.UMat, param: Image2BlobParams = ...) -> cv2.typing.MatLike: ...
+@_typing.overload
+def blobFromImageWithParams(image: cv2.typing.MatLike, blob: cv2.typing.MatLike | None = ..., param: Image2BlobParams = ...) -> cv2.typing.MatLike: ...
+@_typing.overload
+def blobFromImageWithParams(image: cv2.UMat, blob: cv2.UMat | None = ..., param: Image2BlobParams = ...) -> cv2.UMat: ...
+
+@_typing.overload
+def blobFromImages(images: _typing.Sequence[cv2.typing.MatLike], scalefactor: float = ..., size: cv2.typing.Size = ..., mean: cv2.typing.Scalar = ..., swapRB: bool = ..., crop: bool = ..., ddepth: int = ...) -> cv2.typing.MatLike: ...
+@_typing.overload
+def blobFromImages(images: _typing.Sequence[cv2.UMat], scalefactor: float = ..., size: cv2.typing.Size = ..., mean: cv2.typing.Scalar = ..., swapRB: bool = ..., crop: bool = ..., ddepth: int = ...) -> cv2.typing.MatLike: ...
+
+@_typing.overload
+def blobFromImagesWithParams(images: _typing.Sequence[cv2.typing.MatLike], param: Image2BlobParams = ...) -> cv2.typing.MatLike: ...
+@_typing.overload
+def blobFromImagesWithParams(images: _typing.Sequence[cv2.UMat], param: Image2BlobParams = ...) -> cv2.typing.MatLike: ...
+@_typing.overload
+def blobFromImagesWithParams(images: _typing.Sequence[cv2.typing.MatLike], blob: cv2.typing.MatLike | None = ..., param: Image2BlobParams = ...) -> cv2.typing.MatLike: ...
+@_typing.overload
+def blobFromImagesWithParams(images: _typing.Sequence[cv2.UMat], blob: cv2.UMat | None = ..., param: Image2BlobParams = ...) -> cv2.UMat: ...
+
+def getAvailableTargets(be: Backend) -> _typing.Sequence[Target]: ...
+
+@_typing.overload
+def imagesFromBlob(blob_: cv2.typing.MatLike, images_: _typing.Sequence[cv2.typing.MatLike] | None = ...) -> _typing.Sequence[cv2.typing.MatLike]: ...
+@_typing.overload
+def imagesFromBlob(blob_: cv2.typing.MatLike, images_: _typing.Sequence[cv2.UMat] | None = ...) -> _typing.Sequence[cv2.UMat]: ...
+
+@_typing.overload
+def readNet(model: str, config: str = ..., framework: str = ...) -> Net: ...
+@_typing.overload
+def readNet(framework: str, bufferModel: numpy.ndarray[_typing.Any, numpy.dtype[numpy.uint8]], bufferConfig: numpy.ndarray[_typing.Any, numpy.dtype[numpy.uint8]] = ...) -> Net: ...
+
+@_typing.overload
+def readNetFromCaffe(prototxt: str, caffeModel: str = ...) -> Net: ...
+@_typing.overload
+def readNetFromCaffe(bufferProto: numpy.ndarray[_typing.Any, numpy.dtype[numpy.uint8]], bufferModel: numpy.ndarray[_typing.Any, numpy.dtype[numpy.uint8]] = ...) -> Net: ...
+
+@_typing.overload
+def readNetFromDarknet(cfgFile: str, darknetModel: str = ...) -> Net: ...
+@_typing.overload
+def readNetFromDarknet(bufferCfg: numpy.ndarray[_typing.Any, numpy.dtype[numpy.uint8]], bufferModel: numpy.ndarray[_typing.Any, numpy.dtype[numpy.uint8]] = ...) -> Net: ...
+
+@_typing.overload
+def readNetFromModelOptimizer(xml: str, bin: str = ...) -> Net: ...
+@_typing.overload
+def readNetFromModelOptimizer(bufferModelConfig: numpy.ndarray[_typing.Any, numpy.dtype[numpy.uint8]], bufferWeights: numpy.ndarray[_typing.Any, numpy.dtype[numpy.uint8]]) -> Net: ...
+
+@_typing.overload
+def readNetFromONNX(onnxFile: str) -> Net: ...
+@_typing.overload
+def readNetFromONNX(buffer: numpy.ndarray[_typing.Any, numpy.dtype[numpy.uint8]]) -> Net: ...
+
+@_typing.overload
+def readNetFromTFLite(model: str) -> Net: ...
+@_typing.overload
+def readNetFromTFLite(bufferModel: numpy.ndarray[_typing.Any, numpy.dtype[numpy.uint8]]) -> Net: ...
+
+@_typing.overload
+def readNetFromTensorflow(model: str, config: str = ...) -> Net: ...
+@_typing.overload
+def readNetFromTensorflow(bufferModel: numpy.ndarray[_typing.Any, numpy.dtype[numpy.uint8]], bufferConfig: numpy.ndarray[_typing.Any, numpy.dtype[numpy.uint8]] = ...) -> Net: ...
+
+def readNetFromTorch(model: str, isBinary: bool = ..., evaluate: bool = ...) -> Net: ...
+
+def readTensorFromONNX(path: str) -> cv2.typing.MatLike: ...
+
+def readTorchBlob(filename: str, isBinary: bool = ...) -> cv2.typing.MatLike: ...
+
+def shrinkCaffeModel(src: str, dst: str, layersTypes: _typing.Sequence[str] = ...) -> None: ...
+
+def softNMSBoxes(bboxes: _typing.Sequence[cv2.typing.Rect], scores: _typing.Sequence[float], score_threshold: float, nms_threshold: float, top_k: int = ..., sigma: float = ..., method: SoftNMSMethod = ...) -> tuple[_typing.Sequence[float], _typing.Sequence[int]]: ...
+
+def writeTextGraph(model: str, output: str) -> None: ...
+
+

cv2/dnn_superres/__init__.pyi

@@ -0,0 +1,37 @@
+__all__: list[str] = []
+
+import cv2
+import cv2.typing
+import typing as _typing
+
+
+# Classes
+class DnnSuperResImpl:
+    # Functions
+    @classmethod
+    def create(cls) -> DnnSuperResImpl: ...
+
+    def readModel(self, path: str) -> None: ...
+
+    def setModel(self, algo: str, scale: int) -> None: ...
+
+    def setPreferableBackend(self, backendId: int) -> None: ...
+
+    def setPreferableTarget(self, targetId: int) -> None: ...
+
+    @_typing.overload
+    def upsample(self, img: cv2.typing.MatLike, result: cv2.typing.MatLike | None = ...) -> cv2.typing.MatLike: ...
+    @_typing.overload
+    def upsample(self, img: cv2.UMat, result: cv2.UMat | None = ...) -> cv2.UMat: ...
+
+    @_typing.overload
+    def upsampleMultioutput(self, img: cv2.typing.MatLike, imgs_new: _typing.Sequence[cv2.typing.MatLike], scale_factors: _typing.Sequence[int], node_names: _typing.Sequence[str]) -> None: ...
+    @_typing.overload
+    def upsampleMultioutput(self, img: cv2.UMat, imgs_new: _typing.Sequence[cv2.typing.MatLike], scale_factors: _typing.Sequence[int], node_names: _typing.Sequence[str]) -> None: ...
+
+    def getScale(self) -> int: ...
+
+    def getAlgorithm(self) -> str: ...
+
+
+

cv2/dpm/__init__.pyi

@@ -0,0 +1,10 @@
+__all__: list[str] = []
+
+# Classes
+class DPMDetector:
+    # Classes
+    class ObjectDetection:
+        ...
+
+
+

cv2/dynafu/__init__.pyi

@@ -0,0 +1,43 @@
+__all__: list[str] = []
+
+import cv2
+import cv2.kinfu
+import cv2.typing
+import typing as _typing
+
+
+# Classes
+class DynaFu:
+    # Functions
+    @classmethod
+    def create(cls, _params: cv2.kinfu.Params) -> DynaFu: ...
+
+    @_typing.overload
+    def render(self, image: cv2.typing.MatLike | None = ..., cameraPose: cv2.typing.Matx44f = ...) -> cv2.typing.MatLike: ...
+    @_typing.overload
+    def render(self, image: cv2.UMat | None = ..., cameraPose: cv2.typing.Matx44f = ...) -> cv2.UMat: ...
+
+    @_typing.overload
+    def getCloud(self, points: cv2.typing.MatLike | None = ..., normals: cv2.typing.MatLike | None = ...) -> tuple[cv2.typing.MatLike, cv2.typing.MatLike]: ...
+    @_typing.overload
+    def getCloud(self, points: cv2.UMat | None = ..., normals: cv2.UMat | None = ...) -> tuple[cv2.UMat, cv2.UMat]: ...
+
+    @_typing.overload
+    def getPoints(self, points: cv2.typing.MatLike | None = ...) -> cv2.typing.MatLike: ...
+    @_typing.overload
+    def getPoints(self, points: cv2.UMat | None = ...) -> cv2.UMat: ...
+
+    @_typing.overload
+    def getNormals(self, points: cv2.typing.MatLike, normals: cv2.typing.MatLike | None = ...) -> cv2.typing.MatLike: ...
+    @_typing.overload
+    def getNormals(self, points: cv2.UMat, normals: cv2.UMat | None = ...) -> cv2.UMat: ...
+
+    def reset(self) -> None: ...
+
+    @_typing.overload
+    def update(self, depth: cv2.typing.MatLike) -> bool: ...
+    @_typing.overload
+    def update(self, depth: cv2.UMat) -> bool: ...
+
+
+