VRT<T>

Models & Types · AiDotNet.Video.Restoration

VRT: A Video Restoration Transformer for video super-resolution, deblurring, and denoising.

Inheritance: NeuralNetworkBase<T> → VideoNeuralNetworkBase<T> → VideoSuperResolutionBase<T> → VRT<T>

For Beginners

VRT improves video quality by analyzing multiple frames together. Unlike image restoration that processes one frame at a time, VRT uses temporal information to produce better, more consistent results.

Example usage (native mode for training):

Example usage (ONNX mode for inference only):

How It Works

VRT (Video Restoration Transformer) is a powerful architecture for video restoration tasks:

  • Video super-resolution (increasing video resolution)
  • Video deblurring (removing motion blur)
  • Video denoising (removing noise from videos)

Reference: “VRT: A Video Restoration Transformer” https://arxiv.org/abs/2201.12288

Constructors

Constructor Summary
VRT() Initializes a new instance with default architecture settings.
VRT(NeuralNetworkArchitecture<T> architecture, VRTOptions options, IGradientBasedOptimizer<T, Tensor<T>, Tensor<T>> optimizer, ILossFunction<T> lossFunction) Creates a VRT model using native layers for training and inference.
VRT(NeuralNetworkArchitecture<T> architecture, string onnxModelPath, VRTOptions options) Creates a VRT model using a pretrained ONNX model for inference.

Properties

Property Summary
bool SupportsTraining { get; } Gets whether training is supported (only in native mode).

Methods

Method Summary
ModelMetadata<T> GetModelMetadata()
Tensor<T> Deblur(Tensor<T> blurryFrames) Performs video deblurring.
Tensor<T> Denoise(Tensor<T> noisyFrames) Performs video denoising.
Tensor<T> Restore(Tensor<T> input) Restores a video frame using the VRT model.
Tensor<T> SuperResolve(Tensor<T> lowResFrames) Performs video super-resolution.
Tensor<T> Upscale(Tensor<T> lowResFrames)