Partial List of Implemented Layers

Detailed Description

This subsection of dnn module contains information about built-in layers and their descriptions.

Classes listed here, in fact, provides C++ API for creating instances of built-in layers. In addition to this way of layers instantiation, there is a more common factory API (see Utilities for New Layers Registration), it allows to create layers dynamically (by name) and register new ones. You can use both API, but factory API is less convenient for native C++ programming and basically designed for use inside importers (see readNetFromTensorflow()).

Built-in layers reproduce the functionality of the corresponding ONNX and TensorFlow operators. The following layers are among the core building blocks used to assemble imported networks:

• Convolution

• Deconvolution

• Pooling

• InnerProduct

• TanH, ReLU, Sigmoid, BNLL, Power, AbsVal

• Softmax

• Reshape, Flatten, Slice, Split

• LRN

• MVN

• Dropout (since it does nothing on forward pass -))

Classes

Name	Description
class cv::dnn::AbsLayer	View details
class cv::dnn::AccumLayer	View details
class cv::dnn::AcoshLayer	View details
class cv::dnn::AcosLayer	View details
struct cv::dnn::ActivationInt8Params	View details
class cv::dnn::ActivationLayer	View details
class cv::dnn::ActivationLayerInt8	View details
class cv::dnn::AffineGridLayer	View details
class cv::dnn::ArgLayer	ArgMax/ArgMin layer. View details
class cv::dnn::AsinhLayer	View details
class cv::dnn::AsinLayer	View details
class cv::dnn::AtanhLayer	View details
class cv::dnn::AtanLayer	View details
class cv::dnn::AttentionLayer	View details
class cv::dnn::AttentionOnnxAiLayer	View details
class cv::dnn::AveragePoolLayer	View details
class cv::dnn::BaseConvolutionLayer	View details
class cv::dnn::BatchNorm2Layer	View details
class cv::dnn::BatchNormLayer	View details
class cv::dnn::BatchNormLayerInt8	View details
class cv::dnn::BitShiftLayer	View details
class cv::dnn::BlackmanWindowLayer	View details
class cv::dnn::BlankLayer	View details
class cv::dnn::BNLLLayer	View details
class cv::dnn::Cast2Layer	View details
class cv::dnn::CastLayer	View details
class cv::dnn::CeilLayer	View details
class cv::dnn::CeluLayer	View details
class cv::dnn::CenterCropPadLayer	View details
class cv::dnn::ChannelsPReLULayer	View details
class cv::dnn::ClipLayer	View details
class cv::dnn::CompareLayer	View details
class cv::dnn::Concat2Layer	View details
class cv::dnn::ConcatLayer	View details
class cv::dnn::ConstantOfShapeLayer	View details
class cv::dnn::ConstLayer	View details
class cv::dnn::Conv2Int8Layer	View details
struct cv::dnn::Conv2Int8Params	View details
class cv::dnn::Conv2Layer	View details
class cv::dnn::ConvolutionLayer	View details
class cv::dnn::ConvolutionLayerInt8	View details
class cv::dnn::ConvTranspose2Layer	View details
class cv::dnn::CorrelationLayer	View details
class cv::dnn::CoshLayer	View details
class cv::dnn::CosLayer	View details
class cv::dnn::CropAndResizeLayer	View details
class cv::dnn::CropLayer	View details
class cv::dnn::CumSumLayer	View details
class cv::dnn::DataAugmentationLayer	View details
class cv::dnn::DeconvolutionLayer	View details
class cv::dnn::DepthToSpaceLayer	View details
class cv::dnn::DequantizeLayer	View details
class cv::dnn::DequantizeLinearLayer	View details
class cv::dnn::DetectionOutputLayer	Detection output layer. View details
class cv::dnn::DetLayer	View details
class cv::dnn::DFTLayer	View details
class cv::dnn::EinsumLayer	This function performs array summation based on the Einstein summation convention. The function allows for concise expressions of various mathematical operations using subscripts. View details
class cv::dnn::Eltwise2Int8Layer	View details
struct cv::dnn::Eltwise2Int8Params	View details
class cv::dnn::EltwiseLayer	Element wise operation on inputs. View details
class cv::dnn::EltwiseLayerInt8	View details
class cv::dnn::ELULayer	View details
class cv::dnn::ErfLayer	View details
class cv::dnn::Expand2Layer	View details
class cv::dnn::ExpandLayer	View details
class cv::dnn::ExpLayer	View details
class cv::dnn::EyeLikeLayer	View details
class cv::dnn::FlattenLayer	View details
class cv::dnn::FloorLayer	View details
class cv::dnn::FlowWarpLayer	View details
class cv::dnn::Gather2Layer	View details
class cv::dnn::GatherElementsLayer	GatherElements layer GatherElements takes two inputs data and indices of the same rank r >= 1 and an optional attribute axis and works such that: output[i][j][k] = data[index[i][j][k]][j][k] if axis = 0 and r = 3 output[i][j][k] = data[i][index[i][j][k]][k] if axis = 1 and r = 3 output[i][j][k] = data[i][j][index[i][j][k]] if axis = 2 and r = 3. View details
class cv::dnn::GatherLayer	Gather layer. View details
class cv::dnn::GatherNDLayer	GatherND layer. View details
class cv::dnn::GeluApproximationLayer	View details
class cv::dnn::GeluLayer	View details
class cv::dnn::GemmLayer	View details
class cv::dnn::GlobalAveragePoolLayer	View details
class cv::dnn::GridSampleLayer	View details
class cv::dnn::GroupNormLayer	View details
class cv::dnn::GRULayer	GRU recurrent one-layer. View details
class cv::dnn::HammingWindowLayer	View details
class cv::dnn::HannWindowLayer	View details
class cv::dnn::HardmaxLayer	View details
class cv::dnn::HardSigmoidLayer	View details
class cv::dnn::HardSwishLayer	View details
class cv::dnn::IfLayer	View details
struct cv::dnn::InnerProductInt8Params	View details
class cv::dnn::InnerProductLayer	View details
class cv::dnn::InnerProductLayerInt8	View details
class cv::dnn::InstanceNormLayer	View details
class cv::dnn::InterpLayer	Bilinear resize layer from cdmh/deeplab-public-ver2. View details
class cv::dnn::IsInfLayer	View details
class cv::dnn::IsNaNLayer	View details
class cv::dnn::LayerNorm2Layer	View details
class cv::dnn::LayerNormLayer	View details
class cv::dnn::LogLayer	View details
class cv::dnn::LoopLayer	View details
class cv::dnn::LRNLayer	View details
class cv::dnn::LSTM2Layer	View details
class cv::dnn::LSTMLayer	LSTM recurrent layer. View details
class cv::dnn::MatMulInt8Layer	View details
struct cv::dnn::MatMulInt8Params	View details
class cv::dnn::MatMulLayer	View details
class cv::dnn::MaxPoolLayer	View details
class cv::dnn::MaxUnpoolLayer	View details
class cv::dnn::MishLayer	View details
class cv::dnn::MVNLayer	View details
class cv::dnn::NaryEltwiseLayer	View details
class cv::dnn::NegativeLogLikelihoodLossLayer	View details
class cv::dnn::NonMaxSuppressionLayer	View details
class cv::dnn::NonZeroLayer	View details
class cv::dnn::NormalizeBBoxLayer	\( L_p \) - normalization layer. View details
class cv::dnn::NotLayer	View details
class cv::dnn::OneHotLayer	View details
class cv::dnn::Pad2Layer	View details
class cv::dnn::PaddingLayer	Adds extra values for specific axes. View details
class cv::dnn::PermuteLayer	View details
class cv::dnn::Pool2Int8Layer	View details
struct cv::dnn::Pool2Int8Params	View details
class cv::dnn::PoolingLayer	View details
class cv::dnn::PoolingLayerInt8	View details
class cv::dnn::PowerLayer	View details
class cv::dnn::PriorBoxLayer	View details
class cv::dnn::ProposalLayer	View details
class cv::dnn::QuantizeLayer	View details
class cv::dnn::QuantizeLinearLayer	View details
class cv::dnn::RandomNormalLikeLayer	View details
class cv::dnn::RangeLayer	View details
class cv::dnn::ReciprocalLayer	View details
class cv::dnn::Reduce2Layer	View details
class cv::dnn::ReduceLayer	View details
class cv::dnn::RegionLayer	View details
class cv::dnn::ReLU6Layer	View details
class cv::dnn::ReLULayer	View details
class cv::dnn::ReorgLayer	View details
class cv::dnn::RequantizeLayer	View details
class cv::dnn::Reshape2Layer	View details
class cv::dnn::ReshapeLayer	View details
class cv::dnn::Resize2Layer	View details
class cv::dnn::ResizeLayer	Resize input 4-dimensional blob by nearest neighbor or bilinear strategy. View details
class cv::dnn::RMSNormLayer	View details
class cv::dnn::RNNLayer	Classical recurrent layer. View details
class cv::dnn::RoiAlignLayer	View details
class cv::dnn::RotaryEmbeddingLayer	View details
class cv::dnn::RoundLayer	View details
class cv::dnn::ScaleLayer	View details
class cv::dnn::ScaleLayerInt8	View details
class cv::dnn::ScatterLayer	View details
class cv::dnn::ScatterNDLayer	View details
class cv::dnn::SDPALayer	View details
class cv::dnn::SeluLayer	View details
class cv::dnn::ShapeLayer	View details
class cv::dnn::ShiftLayer	View details
class cv::dnn::ShiftLayerInt8	View details
class cv::dnn::ShrinkLayer	View details
class cv::dnn::ShuffleChannelLayer	View details
class cv::dnn::SigmoidLayer	View details
class cv::dnn::SignLayer	View details
class cv::dnn::SinhLayer	View details
class cv::dnn::SinLayer	View details
class cv::dnn::SizeLayer	View details
class cv::dnn::Slice2Layer	View details
class cv::dnn::SliceLayer	View details
class cv::dnn::SoftmaxCrossEntropyLossLayer	View details
class cv::dnn::SoftmaxLayer	View details
class cv::dnn::SoftmaxLayerInt8	View details
class cv::dnn::SoftplusLayer	View details
class cv::dnn::SoftsignLayer	View details
class cv::dnn::SpaceToDepthLayer	View details
class cv::dnn::Split2Layer	View details
class cv::dnn::SplitLayer	View details
class cv::dnn::SqrtLayer	View details
class cv::dnn::SqueezeLayer	View details
class cv::dnn::SwishLayer	View details
class cv::dnn::TanHLayer	View details
class cv::dnn::TanLayer	View details
class cv::dnn::ThresholdedReluLayer	View details
class cv::dnn::Tile2Layer	View details
class cv::dnn::TileLayer	View details
class cv::dnn::TopK2Layer	View details
class cv::dnn::TopKLayer	View details
class cv::dnn::TransformLayoutLayer	View details
class cv::dnn::TransposeLayer	View details
class cv::dnn::TriluLayer	View details
class cv::dnn::UniqueLayer	View details
class cv::dnn::UnsqueezeLayer	View details

Enumerations

enum cv::dnn::ActivationType {
    cv::dnn::ACTIV_NONE = 0,
    cv::dnn::ACTIV_MISH,
    cv::dnn::ACTIV_SWISH,
    cv::dnn::ACTIV_SIGMOID,
    cv::dnn::ACTIV_TANH,
    cv::dnn::ACTIV_ELU,
    cv::dnn::ACTIV_HARDSWISH,
    cv::dnn::ACTIV_HARDSIGMOID,
    cv::dnn::ACTIV_GELU,
    cv::dnn::ACTIV_GELU_APPROX,
    cv::dnn::ACTIV_RELU,
    cv::dnn::ACTIV_CLIP
}

View details

enum cv::dnn::AutoPadding {
    cv::dnn::AUTO_PAD_NONE = 0,
    cv::dnn::AUTO_PAD_SAME_UPPER = 1,
    cv::dnn::AUTO_PAD_SAME_LOWER = 2,
    cv::dnn::AUTO_PAD_VALID = 3
}

View details

enum cv::dnn::LossReduction {
    cv::dnn::LOSS_REDUCTION_NONE = 0,
    cv::dnn::LOSS_REDUCTION_MEAN = 1,
    cv::dnn::LOSS_REDUCTION_SUM = 2
}

View details

Typedef Documentation

ActivationFunc

typedef void(* cv::dnn::ActivationFunc)(const void *input, void *output, size_t len, const float *params)

#include <opencv2/dnn/all_layers.hpp>

Enumeration Type Documentation

ActivationType

enum cv::dnn::ActivationType

#include <opencv2/dnn/all_layers.hpp>

Activation type enumeration for dispatched activation function retrieval.

Enumerator:

ACTIV_NONE Python: cv.dnn.ACTIV_NONE

ACTIV_MISH Python: cv.dnn.ACTIV_MISH

ACTIV_SWISH Python: cv.dnn.ACTIV_SWISH

ACTIV_SIGMOID Python: cv.dnn.ACTIV_SIGMOID

ACTIV_TANH Python: cv.dnn.ACTIV_TANH

ACTIV_ELU Python: cv.dnn.ACTIV_ELU

ACTIV_HARDSWISH Python: cv.dnn.ACTIV_HARDSWISH

ACTIV_HARDSIGMOID Python: cv.dnn.ACTIV_HARDSIGMOID

ACTIV_GELU Python: cv.dnn.ACTIV_GELU

ACTIV_GELU_APPROX Python: cv.dnn.ACTIV_GELU_APPROX

ACTIV_RELU Python: cv.dnn.ACTIV_RELU

ACTIV_CLIP Python: cv.dnn.ACTIV_CLIP

AutoPadding

enum cv::dnn::AutoPadding

#include <opencv2/dnn/all_layers.hpp>

Enumerator:

AUTO_PAD_NONE Python: cv.dnn.AUTO_PAD_NONE

AUTO_PAD_SAME_UPPER Python: cv.dnn.AUTO_PAD_SAME_UPPER

AUTO_PAD_SAME_LOWER Python: cv.dnn.AUTO_PAD_SAME_LOWER

AUTO_PAD_VALID Python: cv.dnn.AUTO_PAD_VALID

LossReduction

enum cv::dnn::LossReduction

#include <opencv2/dnn/all_layers.hpp>

Enumerator:

LOSS_REDUCTION_NONE Python: cv.dnn.LOSS_REDUCTION_NONE

LOSS_REDUCTION_MEAN Python: cv.dnn.LOSS_REDUCTION_MEAN

LOSS_REDUCTION_SUM Python: cv.dnn.LOSS_REDUCTION_SUM

Function Documentation

getActivationFunc()

ActivationFunc cv::dnn::getActivationFunc(int activationType)

#include <opencv2/dnn/all_layers.hpp>

Returns a platform-optimized activation function pointer for the given type. The returned function is selected via CPU dispatch for the best available ISA.