VGG16BNBackbone<T>

Models & Types · AiDotNet.ComputerVision.Detection.Backbones

VGG-16 with batch normalization as a feature-pyramid backbone, cut the way CRAFT cuts it.

Inheritance: NeuralNetworkBase<T> → VGG16BNBackbone<T>

Implements: IDetectionBackbone<T>

How It Works

Thirteen 3x3 convolutions in five blocks (64, 128, 256, 512, 512 channels), each followed by batch norm and ReLU, with 2x2 max pooling after the first four blocks (Simonyan and Zisserman 2015, configuration D). CRAFT (Baek et al. 2019) replaces pool5 and the classifier with fc6/fc7 as convolutions: a 3x3 stride-1 max pool, a 3x3 convolution with dilation 6 to 1024 channels, and a 1x1 convolution to 1024 channels.

Features are returned finest first, [relu2_2, relu3_3, relu4_3, relu5_3, fc7], at strides 2, 4, 8, 16 and 16, the five taps CRAFT’s U-Net decoder merges.

Constructors

Constructor Summary
VGG16BNBackbone(VGG16BNBackboneOptions options) Creates the backbone.

Properties

Property Summary
IReadOnlyList<int> OutputChannels { get; } Channels of each returned feature map, finest first.
IReadOnlyList<int> Strides { get; } Stride of each returned feature map, finest first.
bool IsFrozen { get; set; } Whether the backbone’s weights are frozen.
string Name { get; } The backbone name.

Methods

Method Summary
IFullModel<T, Tensor<T>, Tensor<T>> WithParameters(Vector<T> parameters)
IReadOnlyList<Tensor<T>> GetFeatureMaps(Tensor<T> input)
List<Tensor<T>> ExtractFeatures(Tensor<T> input)
ModelMetadata<T> GetModelMetadata()
ValueTuple<int, int> GetExpectedInputSize() CRAFT’s canonical training resolution.
void Freeze() Freezes the backbone.
void ReadParameters(BinaryReader reader)
void Unfreeze() Unfreezes the backbone.
void WriteParameters(BinaryWriter writer)