RemoteCLIP<T>

Models & Types · AiDotNet.VisionLanguage.Encoders

RemoteCLIP model adapted for remote sensing and satellite imagery understanding.

Inheritance: NeuralNetworkBase<T> → VisionLanguageModelBase<T> → RemoteCLIP<T>

Implements: IContrastiveVisionLanguageModel<T>

For Beginners

RemoteCLIP adapts CLIP for satellite and aerial imagery by fine-tuning on curated remote sensing image-text datasets. It enables zero-shot scene classification of satellite images (e.g., “airport”, “farmland”), image-text retrieval, and object counting in aerial views without task-specific training. Default values follow the original paper settings.

How It Works

RemoteCLIP (Liu et al., 2023) fine-tunes CLIP on curated remote sensing image-text datasets for zero-shot scene classification, image-text retrieval, and object counting in satellite and aerial imagery.

References:

  • Paper: “RemoteCLIP: A Vision Language Foundation Model for Remote Sensing” (Liu et al., 2023)

Example

var architecture = new NeuralNetworkArchitecture<double>(
    inputType: InputType.TwoDimensional,
    taskType: NeuralNetworkTaskType.ImageClassification,
    inputHeight: 224, inputWidth: 224, inputDepth: 3, outputSize: 512);
var trainModel = new RemoteCLIP<double>(architecture, new RemoteCLIPOptions());

Constructors

Constructor Summary
RemoteCLIP(NeuralNetworkArchitecture<T> architecture, RemoteCLIPOptions options, IGradientBasedOptimizer<T, Tensor<T>, Tensor<T>> optimizer)
RemoteCLIP(NeuralNetworkArchitecture<T> architecture, string imageEncoderModelPath, RemoteCLIPOptions options)

Properties

Property Summary
T Temperature { get; }
int EmbeddingDimension { get; }
int MaxSequenceLength { get; }
int ProjectionDimension { get; }
int TextEmbeddingDimension { get; }

Methods

Method Summary
Dictionary<string, T> ZeroShotClassify(Tensor<T> image, string[] labels)
ModelMetadata<T> GetModelMetadata()
T ComputeSimilarity(Tensor<T> image, string text)
Tensor<T> EncodeImage(Tensor<T> image)
Tensor<T> EncodeText(string text)
Tensor<T>[] EncodeTexts(string[] texts)