UniMatch<T>

Models & Types · AiDotNet.Video.Motion

UniMatch unified flow, stereo, and depth estimation with cross-task transfer.

Inheritance: NeuralNetworkBase<T> → VideoNeuralNetworkBase<T> → OpticalFlowBase<T> → UniMatch<T>

For Beginners

UniMatch is a unified model for dense matching that handles optical flow, stereo matching, and depth estimation with a single architecture. It learns general correspondence features.

How It Works

References:

  • Paper: “Unifying Flow, Stereo and Depth Estimation” (Xu et al., TPAMI 2023)

UniMatch unifies optical flow, stereo matching, and depth estimation in a single architecture, enabling cross-task transfer learning.

Example

// Create a UniMatch model for unified flow, stereo, and depth estimation
var uniMatch = new UniMatch<double>();
            
// Or configure with custom parameters
var architecture = new NeuralNetworkArchitecture<double>(
    inputType: InputType.ThreeDimensional,
    taskType: NeuralNetworkTaskType.Regression,
    inputHeight: 256, inputWidth: 256, inputDepth: 3, outputSize: 2);
var model = new UniMatch<double>(architecture);

Constructors

Constructor Summary
UniMatch() Initializes a new instance with default architecture settings.
UniMatch(NeuralNetworkArchitecture<T> architecture, UniMatchOptions options) Creates a new UniMatch model for native training and inference.

Methods

Method Summary
ModelMetadata<T> GetModelMetadata()
Tensor<T> EstimateFlow(Tensor<T> frame0, Tensor<T> frame1)