dnn library

Classes

AsyncArray
Layer
Layer is a wrapper around the cv::dnn::Layer algorithm.
Net
Net allows you to create and manipulate comprehensive artificial neural networks.

Extensions

NetAsync on Net

Constants

DNN_BACKEND_CANN → const int
DNN_BACKEND_CUDA → const int
DNN_BACKEND_DEFAULT → const int
DNN_BACKEND_HALIDE → const int
DNN_BACKEND_INFERENCE_ENGINE → const int
DNN_BACKEND_OPENCV → const int
DNN_BACKEND_TIMVX → const int
DNN_BACKEND_VKCOM → const int
DNN_BACKEND_WEBNN → const int
DNN_TARGET_CPU → const int
DNN_TARGET_CPU_FP16 → const int
Only the ARM platform is supported. Low precision computing, accelerate model inference.
DNN_TARGET_CUDA → const int
DNN_TARGET_CUDA_FP16 → const int
DNN_TARGET_FPGA → const int
FPGA device with CPU fallbacks using Inference Engine's Heterogeneous plugin.
DNN_TARGET_HDDL → const int
DNN_TARGET_MYRIAD → const int
DNN_TARGET_NPU → const int
DNN_TARGET_OPENCL → const int
DNN_TARGET_OPENCL_FP16 → const int
DNN_TARGET_VULKAN → const int

Functions

blobFromImage(InputArray image, {double scalefactor = 1.0, (int, int) size = (0, 0), Scalar? mean, bool swapRB = false, bool crop = false, int ddepth = MatType.CV_32F}) → Mat
Creates 4-dimensional blob from image. Optionally resizes and crops image from center, subtract mean values, scales values by scalefactor, swap Blue and Red channels.
blobFromImageAsync(InputArray image, {double scalefactor = 1.0, (int, int) size = (0, 0), Scalar? mean, bool swapRB = false, bool crop = false, int ddepth = MatType.CV_32F}) → Future<Mat>
blobFromImages(VecMat images, {Mat? blob, double scalefactor = 1.0, (int, int) size = (0, 0), Scalar? mean, bool swapRB = false, bool crop = false, int ddepth = MatType.CV_32F}) → Mat
Creates 4-dimensional blob from series of images. Optionally resizes and crops images from center, subtract mean values, scales values by scalefactor, swap Blue and Red channels. https://docs.opencv.org/4.x/d6/d0f/group__dnn.html#ga0b7b7c3c530b747ef738178835e1e70f
blobFromImagesAsync(VecMat images, {Mat? blob, double scalefactor = 1.0, (int, int) size = (0, 0), Scalar? mean, bool swapRB = false, bool crop = false, int ddepth = MatType.CV_32F}) → Future<Mat>
enableModelDiagnostics(bool isDiagnosticsMode) → void
getAvailableBackends() → List<(int, int)>
getAvailableBackends
getAvailableTargets(int backend) → List<int>
getAvailableTargets https://docs.opencv.org/4.x/d6/d0f/group__dnn.html#ga711e5056b6642b33d9480c98c6889f56
getBlobChannel(Mat blob, int imgidx, int chnidx) → Mat
GetBlobChannel extracts a single (2d)channel from a 4 dimensional blob structure (this might e.g. contain the results of a SSD or YOLO detection,
getBlobChannelAsync(Mat blob, int imgidx, int chnidx) → Future<Mat>
getBlobSize(Mat blob) → VecI32
GetBlobSize retrieves the 4 dimensional size information in (N,C,H,W) order
imagesFromBlob(Mat blob) → List<Mat>
ImagesFromBlob Parse a 4D blob and output the images it contains as 2D arrays through a simpler data structure (std::vectorcv::Mat).
imagesFromBlobAsync(Mat blob) → Future<List<Mat>>
NMSBoxes(VecRect bboxes, VecF32 scores, double scoreThreshold, double nmsThreshold, {double eta = 1.0, int topK = 0}) → List<int>
NMSBoxes performs non maximum suppression given boxes and corresponding scores.
NMSBoxesAsync(VecRect bboxes, VecF32 scores, double scoreThreshold, double nmsThreshold, {double eta = 1.0, int topK = 0}) → Future<List<int>>