Computer Vision Models (Vision API)

Native Lux Models

Boltz.Vision.AlexNet Type

AlexNet(; pretrained=false)

Create an AlexNet model [4].

Keyword Arguments

• pretrained: Valid Options are false, true, :DEFAULT, :ImageNet1K or :ImageNet1K_V1. :DEFAULT, true and :ImageNet1K weights currently corresponds to :ImageNet1K_V1.

source

Boltz.Vision.EfficientNet Type

EfficientNet(variant::Union{String,Symbol}; pretrained::Bool=false, in_channels::Int=3,
             nclasses::Int=1000)

Create an EfficientNet model [5].

Arguments

• variant::Union{String,Symbol}: The variant of the EfficientNet model to use. Valid variants are :b0, :b1, :b2, :b3, :b4, :b5, :b6, :b7.

Keyword Arguments

• pretrained::Bool=false: If true, loads pretrained weights when LuxCore.setup is called.

• in_channels::Int=3: The number of input channels. (Must be 3 for pretrained models.)

• nclasses::Int=1000: The number of output classes. (Must be 1000 for pretrained models.)

source

Boltz.Vision.VGG Type

VGG(imsize; config, inchannels, batchnorm = false, nclasses, fcsize, dropout)

Create a VGG model [6].

Arguments

• imsize: input image width and height as a tuple

• config: the configuration for the convolution layers

• inchannels: number of input channels

• batchnorm: set to true to use batch normalization after each convolution

• nclasses: number of output classes

• fcsize: intermediate fully connected layer size

• dropout: dropout level between fully connected layers

source

VGG(depth::Int; batchnorm::Bool=false, pretrained::Bool=false)

Create a VGG model [6] with ImageNet Configuration.

Arguments

• depth::Int: the depth of the VGG model. Choices: {11, 13, 16, 19}.

Keyword Arguments

• batchnorm = false: set to true to use batch normalization after each convolution.

• pretrained::Bool=false: If true, loads pretrained weights when LuxCore.setup is called.

source

Boltz.Vision.VisionTransformer Type

VisionTransformer(name::Symbol; pretrained=false)

Creates a Vision Transformer model with the specified configuration.

Arguments

• name::Symbol: name of the Vision Transformer model to create. The following models are available – :tiny, :small, :base, :large, :huge, :giant, :gigantic.

Keyword Arguments

• pretrained::Bool=false: If true, loads pretrained weights when LuxCore.setup is called.

source

Imported from Metalhead.jl

Load Metalhead

You need to load Metalhead before using these models.

Boltz.Vision.ConvMixer Function

ConvMixer(name::Symbol; pretrained::Bool=false)

Create a ConvMixer model [7].

Arguments

• name::Symbol: The name of the ConvMixer model. Must be one of :base, :small, or :large.

Keyword Arguments

• pretrained::Bool=false: If true, loads pretrained weights when LuxCore.setup is called.

source

Boltz.Vision.DenseNet Function

DenseNet(depth::Int; pretrained::Bool=false)

Create a DenseNet model [8].

Arguments

• depth::Int: The depth of the DenseNet model. Must be one of 121, 161, 169, or 201.

Keyword Arguments

• pretrained::Bool=false: If true, loads pretrained weights when LuxCore.setup is called.

source

Boltz.Vision.GoogLeNet Function

GoogLeNet(; pretrained::Bool=false)

Create a GoogLeNet model [9].

Keyword Arguments

• pretrained::Bool=false: If true, loads pretrained weights when LuxCore.setup is called.

source

Boltz.Vision.MobileNet Function

MobileNet(name::Symbol; pretrained::Bool=false)

Create a MobileNet model [10–12].

Arguments

• name::Symbol: The name of the MobileNet model. Must be one of :v1, :v2, :v3_small, or :v3_large.

Keyword Arguments

• pretrained::Bool=false: If true, loads pretrained weights when LuxCore.setup is called.

source

Boltz.Vision.ResNet Function

ResNet(depth::Int; pretrained::Bool=false)

Create a ResNet model [13].

Arguments

• depth::Int: The depth of the ResNet model. Must be one of 18, 34, 50, 101, or 152.

Keyword Arguments

• pretrained::Bool=false: If true, loads pretrained weights when LuxCore.setup is called.

source

Boltz.Vision.ResNeXt Function

ResNeXt(depth::Int; cardinality=32, base_width=nothing, pretrained::Bool=false)

Create a ResNeXt model [14].

Arguments

• depth::Int: The depth of the ResNeXt model. Must be one of 50, 101, or 152.

Keyword Arguments

• pretrained::Bool=false: If true, loads pretrained weights when LuxCore.setup is called.

• cardinality: The cardinality of the ResNeXt model. Defaults to 32.

• base_width: The base width of the ResNeXt model. Defaults to 8 for depth 101 and 4 otherwise.

source

Boltz.Vision.SqueezeNet Function

SqueezeNet(; pretrained::Bool=false)

Create a SqueezeNet model [15].

Keyword Arguments

• pretrained::Bool=false: If true, loads pretrained weights when LuxCore.setup is called.

source

Boltz.Vision.WideResNet Function

WideResNet(depth::Int; pretrained::Bool=false)

Create a WideResNet model [16].

Arguments

• depth::Int: The depth of the WideResNet model. Must be one of 18, 34, 50, 101, or 152.

Keyword Arguments

• pretrained::Bool=false: If true, loads pretrained weights when LuxCore.setup is called.

source

Pretrained Models (ImageNet1K)

Load Pretrained Weights

Pass pretrained=true to the model constructor to load the pretrained weights.

MODEL	Additional Packages	TOP 1 ACCURACY (%)	TOP 5 ACCURACY (%)
AlexNet()	JLD2	54.48	77.72
VGG(11)	JLD2	67.35	87.91
VGG(13)	JLD2	68.40	88.48
VGG(16)	JLD2	70.24	89.80
VGG(19)	JLD2	71.09	90.27
VGG(11; batchnorm=true)	JLD2	69.09	88.94
VGG(13; batchnorm=true)	JLD2	69.66	89.49
VGG(16; batchnorm=true)	JLD2	72.11	91.02
VGG(19; batchnorm=true)	JLD2	72.95	91.32
EfficientNet(:b0)	Pickle	-	-
EfficientNet(:b1)	Pickle	-	-
EfficientNet(:b2)	Pickle	-	-
EfficientNet(:b3)	Pickle	-	-
EfficientNet(:b4)	Pickle	-	-
EfficientNet(:b5)	Pickle	-	-
EfficientNet(:b6)	Pickle	-	-
EfficientNet(:b7)	Pickle	-	-
ResNet(18)	JLD2, Metalhead	-	-
ResNet(34)	JLD2, Metalhead	-	-
ResNet(50)	JLD2, Metalhead	-	-
ResNet(101)	JLD2, Metalhead	-	-
ResNet(152)	JLD2, Metalhead	-	-
ResNeXt(50; cardinality=32, base_width=4)	JLD2, Metalhead	-	-
ResNeXt(101; cardinality=32, base_width=8)	JLD2, Metalhead	-	-
ResNeXt(101; cardinality=64, base_width=4)	JLD2, Metalhead	-	-
SqueezeNet()	JLD2, Metalhead	-	-
WideResNet(50)	JLD2, Metalhead	-	-
WideResNet(101)	JLD2, Metalhead	-	-

Pretrained Models from Metalhead

For Models imported from Metalhead, the pretrained weights can be loaded if they are available in Metalhead. Refer to the Metalhead.jl docs for a list of available pretrained models.

Preprocessing

All the pretrained models require that the images be normalized with the parameters mean = [0.485f0, 0.456f0, 0.406f0] and std = [0.229f0, 0.224f0, 0.225f0].