SkyEyeGPT<T>

Models & Types · AiDotNet.VisionLanguage.RemoteSensing

SkyEyeGPT: unified remote sensing vision-language tasks with 968K instruction samples.

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

Implements: IRemoteSensingVLM<T>

For Beginners

SkyEyeGPT is a vision-language model specialized for satellite and aerial imagery understanding tasks. Default values follow the original paper settings.

How It Works

SkyEyeGPT (Zhan et al., 2024) unifies remote sensing vision-language tasks via instruction tuning with a large language model. Trained on 968K instruction samples covering captioning, visual question answering, grounding, and scene classification, it adapts a general-purpose VLM to the remote sensing domain with satellite and aerial imagery understanding.

References:

  • Paper: “SkyEyeGPT: Unifying Remote Sensing Vision-Language Tasks via Instruction Tuning with Large Language Model (Zhan et al., 2024)”

Example

// Create a SkyEyeGPT model for unified remote sensing vision-language tasks
// with 968K instruction samples for captioning, VQA, and grounding
var architecture = new NeuralNetworkArchitecture<double>(
    inputType: InputType.TwoDimensional,
    taskType: NeuralNetworkTaskType.ImageClassification,
    inputHeight: 224, inputWidth: 224, inputDepth: 3, outputSize: 512);
            
// ONNX inference mode with pre-trained model
var model = new SkyEyeGPT<double>(architecture, "skyeyegpt.onnx");
            
// Training mode with native layers
var trainModel = new SkyEyeGPT<double>(architecture, new SkyEyeGPTOptions());

Constructors

Constructor Summary
SkyEyeGPT(NeuralNetworkArchitecture<T> architecture, SkyEyeGPTOptions options, IGradientBasedOptimizer<T, Tensor<T>, Tensor<T>> optimizer)
SkyEyeGPT(NeuralNetworkArchitecture<T> architecture, string modelPath, SkyEyeGPTOptions options)

Properties

Property Summary
int DecoderEmbeddingDim { get; }
int EmbeddingDimension { get; }
int MaxGenerationLength { get; }
string LanguageModelName { get; }
string SupportedBands { get; }

Methods

Method Summary
ModelMetadata<T> GetModelMetadata()
Tensor<T> AnswerRemoteSensingQuestion(Tensor<T> image, string question)
Tensor<T> EncodeImage(Tensor<T> image)
Tensor<T> GenerateFromImage(Tensor<T> image, string prompt) Generates text from a remote sensing image using SkyEyeGPT’s unified multi-task pipeline.