# Herculaneum Scroll · ML Pipeline # WEDS Expedition · Vesuvius Challenge import numpy as np import torch from weds.vision import PhaseContrastCT class ScrollSegmenter(torch.nn.Module): def __init__(self, resolution=8): self.voxel_res = resolution # μm self.layers = UNet3D(depth=5) self.decoder = InkDetector() def forward(self, x): x = self.layers(x) return self.decoder.predict_ink(x) # Load 8µm tomographic slices scan = PhaseContrastCT.load( "PHerc1667_segment_1.tif" ) model = ScrollSegmenter(resolution=8) text = model.forward(scan.voxels) print(f"Recovered: {text.confidence:.1%}") # → Recovered: 94.3%
Applying machine learning, remote sensing, and computational analysis to the world's most demanding scientific and artistic field expeditions.
We sit at the convergence of rigorous scientific fieldwork and advanced computational methods — where satellite imagery meets papyrus, where sonar maps ancient seafloors.
3D computed tomography, multi-spectral imaging and neural ink detection applied to carbonised papyri from Herculaneum, fragments from the Dead Sea and beyond.
Sonar bathymetry, ROV photogrammetry and deep-learning wreck classification to map and document submerged cultural heritage and geological formations.
Processing satellite, ice-core and ocean sensor data to reconstruct paleoclimate records, model future extremes, and support scientific expeditions in polar and arid environments.
LiDAR, hyperspectral and SAR imaging to discover hidden landscapes — from buried Roman roads to subglacial rivers beneath Antarctic ice sheets.
eDNA sequencing, acoustic monitoring and population modelling in biodiversity-rich expedition environments, from deep ocean trenches to tropical canopy.
Integrated data platforms for expedition planning, real-time field analytics, and post-expedition archival — built to RGS and BS 8848 standards.
From the Mediterranean seabed to Mount Vesuvius, our collective deploys data science where it has never been applied before.
Working alongside the Vesuvius Challenge, applying 3D convolutional neural networks to phase-contrast X-ray CT scans of carbonised papyri buried by the AD 79 eruption. Our pipeline detects trace carbon ink at sub-voxel resolution — recovering text lost for two millennia from the Villa of the Papyri library.
Re-examining the original 1872–76 sounding data from the Challenger Expedition using modern interpolation and deep-learning bathymetry classification to build the highest-resolution historic seafloor map ever produced.
Processing multi-proxy paleoclimate datasets from Antarctic and Greenland ice cores. Machine learning models reconstruct 800,000-year temperature records and identify climate tipping-point signatures invisible to classical statistical methods.
World Exploration Data Science was founded by Matthew Marriott FRAS FRGS — explorer, writer, data scientist, and Fellow of both the Royal Asiatic Society and the Royal Geographical Society. Drawing on a lifetime of fieldwork across deserts, oceans, and ancient civilisations, Matthew established WEDS to close the gap between cutting-edge computational methods and the reality of scientific and artistic expeditions in the field.
Matthew's heritage runs deep through exploration history. He is a direct descendant of Sir John Murray, the pioneering oceanographer of the HMS Challenger Expedition (1872–76) — the voyage that founded the science of oceanography and first mapped the deep ocean floor. That tradition of rigorous, evidence-based field science is the founding DNA of WEDS.
A published author (The Renaissance Diet: Eat and Drink Like Leonardo Da Vinci), landscape photographer, and host of the Whiskey and a Map: True Stories of Adventure podcast, Matthew brings the perspective of a storyteller as well as a scientist to the collective's work — ensuring that data captured in the field is not only analytically rigorous but compellingly communicated.
WEDS draws together data scientists, oceanographers, archaeologists, climatologists, and field explorers united by a single conviction: the most important data in the world is still waiting to be found.
Enquiries about collaboration, expedition partnership, or data science commissions.
Enquiries about collaboration, expedition partnership, or data science commissions are welcome.
✉ Contact UsWe offer specialist consultation in exploration data science — from expedition field analytics and remote sensing pipeline design, to ML-powered artefact recovery and climate data modelling. Whether you are a scientific institution, documentary team, cultural heritage body, or independent expedition, we bring rigorous data science to the places that need it most.