MAPL-EMIT, an end-to-end vision transformer, detects 79% of known hand-annotated EMIT methane plume complexes and twice as many plausible plumes as human analysts while supporting quantification and multi-plume handling.
The Pearson correlation coefficient is computed across overlapping pixels, requiring a correlation exceeding 0.97 for the candidates to be classified as identical
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Global monitoring of methane point sources using deep learning on hyperspectral radiance measurements from EMIT
MAPL-EMIT, an end-to-end vision transformer, detects 79% of known hand-annotated EMIT methane plume complexes and twice as many plausible plumes as human analysts while supporting quantification and multi-plume handling.