C2fBlock<T>
Models & Types · AiDotNet.ComputerVision.Detection.Backbones
YOLOv8’s C2f (Jocher et al. 2023): split into two halves, run n bottlenecks on the second, and concatenate both halves with EVERY bottleneck’s output before a 1x1 fusion Conv.
Inheritance: LayerBase<T> → C2fBlock<T>
Implements: IShapeContract
How It Works
The reference cv1 is one 1x1 Conv to 2c channels whose output is chunked in two; two 1x1 Convs to c each compute exactly the same function (the weight rows partition), without a channel slice.
Constructors
| Constructor | Summary |
|---|---|
C2fBlock(int outChannels, int depth, bool shortcut) |
Creates a C2f of outChannels with depth bottlenecks. |
Properties
| Property | Summary |
|---|---|
bool SupportsTraining { get; } |
Methods
| Method | Summary |
|---|---|
IReadOnlyList<ILayer<T>> GetSubLayers() |
|
IReadOnlyList<OutputAxisContract> OutputAxesFor(int inputRank) |
|
void ResetState() |
|
void SetTrainingMode(bool isTraining) |
|
void UpdateParameters(T learningRate) |