Map what no one has mapped
LiDAR drones build 3D models of mines, tunnels and abandoned workings where GPS cannot reach, turning unknown voids into measured plans.
Like a CT reconstructionThe imaging AI that finds tumors and hairline fractures, pointed at satellite and drone data. Find critical minerals, map tunnels and catch fractures before they fail.
RadSherpa medical AI · used in 89+ countries · 2,700+ daily users
Pit wall · tunnel roof · X-ray · rescue
RadSherpa's diagnostic AI already helps radiologists read X-rays, mammograms, CT and ultrasound around the world. MineSherpa points that same technology at the ground.
Country, daily-user, image and experience figures are as published by RadSherpa on radsherpa.com (home, About and Team pages, reviewed 28 September 2026). They describe RadSherpa's medical imaging platform, not mining deployments. Mining applications on this site are proposed extensions of the same techniques.
Medical imaging AI is trained to find something small and faint inside a huge, noisy image, measure it, track it over time and flag it for a specialist before it becomes a crisis. That is exactly the job in mining.
A few millimetres of abnormal tissue hidden among millions of pixels.
A faint alteration signature in satellite and hyperspectral imagery across thousands of square kilometres.
The thin line a tired eye misses and a trained model flags.
Cracks and joint sets in drone imagery, found before a wedge or slab lets go.
Segment it, measure it in 3D, compare it with the last scan.
Drone photogrammetry, LiDAR and spectral mapping turn a target into tonnes, volumes and grade zones.
Is it growing? Did treatment work? Change detection answers.
Millimetre-scale ground movement and new cracking around slopes, dams and tailings — caught early.
AI proposes, a specialist confirms, every finding is audited.
Every flag goes to your qualified person with its evidence, confidence and date.
“If it’s good enough for medicine, it’s good enough for your mine.”Pradeep Albert, MD · Founder, RadSherpa
Pick an image and run detection. The same filter traces a hairline fracture on an X-ray and open cracks in a pit wall, then measures each one — right here in your browser.
Images are synthetic and generated in your browser, so detections can be scored against a known answer. Detection runs live using a multi-scale ridge filter, the classical building block of fracture and vessel detection in medical imaging, as a stand-in for RadSherpa's trained models. Risk flags are prompts for geotechnical review, not stability assessments.
Mines, tunnels and underground facilities are dark, unmapped and unstable. RadSherpa turns drone and sensor data into the same kind of read a trauma team gets from a CT scan: what is there, what is broken, and where the people are.
LiDAR drones build 3D models of mines, tunnels and abandoned workings where GPS cannot reach, turning unknown voids into measured plans.
Like a CT reconstructionThermal and visual detection locates trapped or missing workers after a collapse, fire or flood, so rescue teams know where to go before they go in.
Like finding a noduleFracture, roof-sag and deformation detection flags unstable ground in tunnels, shafts and portals, and compares each flight with the last.
Like a fracture readSatellite radar and imagery track surface subsidence, new excavation and water over entire regions, around the clock and through cloud.
Like population screeningSix jobs, one imaging brain. Each one borrows a proven technique from medicine.
Satellites screen whole regions for mineral, water and vegetation signals. Drones zoom into the targets at centimetre scale. MineSherpa is designed to fuse both with field samples into one evidence trail.
Multispectral and hyperspectral scenes flag alteration minerals, vegetation stress, surface water and land change across whole districts.
SAR sees through cloud and at night; InSAR time series can reveal millimetre-scale ground movement around pits, dams and tailings.
Photogrammetry, LiDAR, multispectral and thermal payloads map shortlisted areas at centimetre resolution, including terrain under vegetation.
Every anomaly is tied to sample IDs, coordinates and accredited laboratory assays. Spectral signals are leads, never grades.
Models rank targets and flag changes; qualified reviewers sign off. The export keeps sensor, date and uncertainty for every layer.
Proposed capability. The viewer is procedurally generated artwork; no satellite or drone data is connected and no site has been analyzed. Resolution figures are typical published ranges for public missions and commercial drone payloads. Remote-sensing signals require field sampling, laboratory assays and qualified review before any resource, safety or investment conclusion.
Early detection pays three ways: fewer incidents and stoppages, fewer wasted drill holes, and less time spent on manual inspection. Put in your own numbers.
The starting values are example inputs, not benchmarks or promises. Replace them with your operation's figures. Results are simple arithmetic on your assumptions and exclude program costs.
Dr. Pradeep Albert is a board-certified radiologist with more than 20 years in medical imaging. He founded RadSherpa to help radiologists read faster and more accurately. Radiology and mineral exploration solve the same problem: find subtle anomalies in noisy, multi-channel images, measure them honestly and route them to a specialist. RadSherpa now brings that discipline to mining.
Founder, RadSherpa · Board-certified radiologist
Ask how medical imaging AI applies to your mine, tunnel or deposit — or about any operation, commodity, price or earthquake in the data hub.
USGS estimates 2025 world mine production at about 110,000 t, with China at 40,000 t, Russia 32,000 t and Tajikistan 22,000 t. U.S. net import reliance is 91%.
Answers come from this site's sourced data. Open-ended AI on your own data runs in a RadSherpa engagement.
Drop in a technical report or data file. It stays in your browser until you choose to share it.
Explore technical, environmental, infrastructure, and commercial questions in the same workspace.
Explore the project mapExport a sourced brief: what is supported, what is uncertain, and what needs a qualified person's review.
Send one drone survey, one tunnel or one area of interest. We'll show you what medical-grade imaging AI sees in it.
RadSherpa's imaging runs on context: where the world's major mines are, which minerals are in short supply, what prices and events are moving the market. It's all here.
A WORLD OF
POSSIBILITY.
ONE INTELLIGENCE LAYER.
Operations are compiled from public sources (company sites, filings, Wikipedia, USGS MRDS, OpenStreetMap); each profile lists its sources, the date of ownership information and how precise its map point is. Earthquakes are live from the USGS M4.5+ weekly feed. The antimony case-study layer is labelled separately. No ownership, partnership or endorsement is implied.
Share of world mine production held by the largest producer and the top three, with U.S. net import reliance — the starting point for any national critical-minerals strategy.
Shares are calculated from USGS Mineral Commodity Summaries 2026 estimates for 2025 (uranium: World Nuclear Association, 2024). Lithium and antimony world totals exclude withheld U.S. production. PGMs show platinum. Mine production only; refining is often more concentrated. Click a row for the full profile and sources.
Combine geological survey records, concession registers and satellite screening into one map of what a country holds and what is still unexplored.
Track which critical minerals depend on one or two supplier countries, and where new domestic or allied supply could come from.
Use satellite change detection and InSAR to monitor permitted sites, illegal mining, tailings dams and rehabilitation commitments.
Every figure carries its source and date, so briefings to ministers and investors stand up to scrutiny.
Prices: indicative spot quotes from gold-api.com, refreshed every 60 seconds; arrows and trend lines compare with earlier refreshes in this browser session, not a daily close. Lithium, nickel, cobalt, rare earths, graphite and antimony have no free public spot feed, so no price is shown. Headline counts come from the GDELT news index for the past 7 days. Not investment advice.
The back-of-envelope maths every geologist, analyst and buyer does daily, wired to the live spot feed.
Statistics, markets, reporting standards, filings, geoscience, imagery and safety — the places professionals check first.
Explore Bolivia and Colombia through the evidence: what a company disclosed, what a source supports, and what still needs verification.
Bolivia × Colombia
Review date: 28 Sep 2026Try a project name, a country, or reset the filters.
Keep a dated claim from becoming an unsupported conclusion.
CleanTech’s April 2025 release cites roughly 3,000 tonnes per year and 3.6% of global production. These are dated, company-cited figures, not a verified current USGS estimate. A like-for-like year and denominator should be confirmed before updating a national production statistic.
UAMY’s filing for the quarter ended 31 March 2026 describes antimony processing in Montana and Mexico. The reviewed filing does not substantiate the supplied claim of a UAMY-owned hydrometallurgical facility in Bolivia. Do not publish that facility claim as established fact.
This review establishes neither a commercial Colombian antimony mine nor a national zero-production estimate. Minastyc is retained as broader critical-minerals context; it should not be relabeled an antimony producer without mineral-specific evidence.
Proposed scope: align drone imagery, survey control and an authorized site boundary in one coordinate system. Keep sensor, date and processing records with each layer.
Proposed scope: attach sample IDs, laboratory results and field observations to candidate features. Do not convert a color anomaly into a verified antimony grade.
Proposed scope: separate observed changes from model interpretations, document uncertainty and route geological or engineering conclusions to qualified reviewers.
No site imagery has been analyzed for these profiles. No grade, stockpile volume, slope safety or operational decision is certified by this preview.
Sources were inspected for attribution and scope on 28 September 2026. Company disclosures remain attributed to the issuer. This was not a field investigation, title verification or exhaustive review of later records. No current USGS production statistic is claimed: the linked 2026 table was unavailable during review. Third-party mine-index links below are user-submitted leads, not verified primary records.
Use “Show on map” on any profile to open its labeled regional locator. Independent third-party projects and sources. No ownership, partnership, commercial relationship, endorsement or verified operating status is implied for MineSherpa, RadSherpa or America First Global.