PANet<T>
Models & Types · AiDotNet.ComputerVision.Detection.Necks
Path Aggregation Network (PANet) for enhanced multi-scale feature fusion.
Inheritance: ModelBase<T, Tensor<T>, Tensor<T>> → NeckBase<T> → PANet<T>
For Beginners
PANet improves upon FPN by adding a bottom-up pathway after the top-down pathway. This creates a bidirectional flow of information, allowing both high-level semantics to flow down and low-level details to flow up.
How It Works
Key features:
- FPN-style top-down pathway
- Additional bottom-up pathway for better localization
- Used in YOLOv4, YOLOv5, and many modern detectors
Reference: Liu et al., “Path Aggregation Network for Instance Segmentation”, CVPR 2018
Constructors
| Constructor | Summary |
|---|---|
PANet(NeckConfig config) |
Creates PANet from a configuration object. |
PANet(int[] inputChannels, int outputChannels) |
Creates a new Path Aggregation 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) |