VideoTransformer class

A Transformer model adapted for video classification (e.g., action recognition).

This model expects pre-extracted video frame or clip embeddings as a sequence of ValueVectors. These embeddings could come from a pre-trained CNN (like ResNet) or a Vision Transformer applied per frame.

Inheritance

• Object
• Module
• VideoTransformer

Constructors

VideoTransformer({required int frameEmbedDim, required int embedSize, required int maxVideoSequenceLength, required int numClasses, int numLayers = 2, int numHeads = 4})

Properties

embedSize → int

final

frameEmbedDim → int

final

frameProjection → Layer?

final

hashCode → int

The hash code for this object.

no setter, inherited

maxVideoSequenceLength → int

final

mlpHead → Layer

final

numClasses → int

final

numHeads → int

final

numLayers → int

final

positionEmbeddings → List<ValueVector>

final

runtimeType → Type

A representation of the runtime type of the object.

no setter, inherited

transformerEncoder → TransformerEncoder

final

Methods

forward(List<ValueVector> videoEmbeddings) → List<Value>

The forward pass for the Video Transformer.

noSuchMethod(Invocation invocation) → dynamic

Invoked when a nonexistent method or property is accessed.

inherited

parameters() → List<Value>

override

toString() → String

A string representation of this object.

inherited

zeroGrad() → void

inherited

Operators

operator ==(Object other) → bool

The equality operator.

inherited