FPN<T>
Models & Types · AiDotNet.ComputerVision.Detection.Necks
Feature Pyramid Network (FPN) for multi-scale feature fusion.
Inheritance: ModelBase<T, Tensor<T>, Tensor<T>> → NeckBase<T> → FPN<T>
For Beginners
FPN creates a feature pyramid by combining features from different backbone levels using a top-down pathway. High-resolution features from earlier layers are enriched with semantic information from deeper layers.
How It Works
Key features:
- Top-down pathway with lateral connections
- 1x1 lateral convolutions to match channel dimensions
- Simple element-wise addition for fusion
- A 3x3 convolution on each merged map, with no nonlinearity (Lin et al. 2017, Sec. 3)
- Fast and memory efficient
Reference: Lin et al., “Feature Pyramid Networks for Object Detection”, CVPR 2017
Constructors
| Constructor | Summary |
|---|---|
FPN(NeckConfig config) |
Creates FPN from a configuration object. |
FPN(int[] inputChannels, int outputChannels) |
Creates a new Feature Pyramid Network. |
Properties
| Property | Summary |
|---|---|
int NumLevels { get; } |
|
int OutputChannels { get; } |
|
string Name { get; } |
Methods
| Method | Summary |
|---|---|
List<Tensor<T>> Forward(List<Tensor<T>> features) |
|
long GetParameterCount() |
|
void ReadParameters(BinaryReader reader) |
|
void WriteParameters(BinaryWriter writer) |