Text to Tensor 📜➡️🖼️

The task of TextToTensor involves the transforms textual data into a tensor representation, assisting in the integration of natural language processing in the EO domain.

📌 Inputs:

• TextPrompt [input]: - Textual data or queries for tensor conversion.
• SampleTensor [target]: - A tensor ranging from 1D to 5D.
• SampleTensor [extra]: - Additional tensors that may assist in exploratory data analysis or the training process.

🛠️ Use Cases:

• Research Linking: Connect satellite images with associated environmental, urban, or geological studies by referencing the documentation.

• Interactive Geospatial Querying: Enable researchers and analysts to directly locate areas of interest in large satellite datasets using textual descriptions.

• Rapid Data Highlighting: Visualize specific regions within geospatial datasets based on current events, such as areas affected by recent natural disasters, deforestation, or urban expansion.

🔍 Example:

import mlstac

name = "https://huggingface.co/datasets/mlstac/text_to_tensor_demo.json"
dataset = mlstac.dataset(name, streaming=True, framework="torch")
print(next(iter(dataset)))

# {'input': 'hola que hace', 'target': tensor([[[[...]]]]))}