Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
October 2, 2025Open Access

TESSERA: Precomputed FAIR Global Pixel Embeddings for Earth Representation and Analysis

View Full Paper
Ask AI
Bookmark
Share

Authors

ZFZhengpeng FengCAClement AtzbergerSJSadiq Jaffer

Discussion

Loading...

Member takes

Overview

Foundation model develops 128-dimensional embeddings for Earth Observation, suggesting enhanced accuracy and ease of use.

Key Points

  • Tessera achieves state-of-the-art performance across diverse complex tasks in Earth Observation.
  • The model creates 128-dimensional latent embeddings requiring few labeled examples to train effectively.
  • It combines optical data with synthetic aperture radar backscatter for enhanced data quality at 10m resolution.
  • Tessera provides unprecedented global, annual embedding maps that outperform existing task-specific models.

Cite This Study

Feng et al. (2025) studied this question.

synapsesocial.com/papers/68de5d9c83cbc991d0a2051ehttps://doi.org/10.48550/arxiv.2506.20380
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1TESSERA: Temporal Embeddings of Surface Spectra for Earth Representation and Analysis2025 · 13 citations
  2. 2TESSERA: Temporal Embeddings of Surface Spectra for Earth Representation and Analysis2025 · 10 citations
  3. 3Democratizing planetary-scale analysis: an ultra-lightweight Earth embedding database for accurate and flexible global land monitoring2026 · 1 citations
  4. 4Evaluating AlphaEarth and TESSERA Geospatial Embeddings for Machine Learning-Based Burned Area Mapping in Portugal2026
  5. 5Producing global cloud-less Landsat monthly time series 2000-2025 using iterative aggregation and gap-filling2026