From Scanned PDFs to
Structured Data.
Drag. Drop. Download.
NDepth extracts structured tabular data from scanned core analysis reports, mud logs, and subsurface PDFs. AI reads handwritten notes, faded tables, and complex multi-column layouts with high accuracy.
Specialized AI for Every Document
Specialized AI models trained on thousands of geological documents for maximum accuracy.
Extract Porosity, Permeability & More
Digitize decades of core analysis reports. Our AI reads handwritten notes, faded tables, and complex multi-column layouts with high accuracy.
Hydrocarbon Show Extraction
Pull gas readings, show descriptions, and formation tops from mud logs - even from the messiest handwritten field reports.
How NDepth Works
Every document goes through a six-stage pipeline. AI handles the heavy lifting - you handle the final QC.
OCR & Enhancement
Advanced OCR with auto-rotation, deskew, contrast enhancement, and column detection. Low-confidence pages are automatically re-processed with enhanced imaging.
Structure Classification
Every page is classified by visual structure: tabular data, graphical logs, maps, photographs, or forms. Tables always take priority over other content types.
Table Type Classification
Tabular pages are classified into dozens of specific types using column fingerprinting, keyword matching, and header analysis. Each type gets a specialized extraction model.
Intelligent Extraction
Data is extracted with column alignment validation, depth continuity logic, ditto mark resolution, and precise handling of multi-page tables that span across pages.
Duplicate Detection
Multi-signal dedup using Jaccard similarity on headers, sample number overlap analysis, depth range intersection, and continuation detection to prevent data duplication.
Data Validation & QC
Values are validated against geological ranges (porosity 0-40%, grain density 2.0-3.5 g/cc). OCR errors are auto-corrected and out-of-range values are flagged for review.
Every Field, Automatically
NDepth identifies and standardizes well metadata from headers, cover pages, and document body text.
Standardized to 14 digits with state code validation
Parsed from headers, cover pages, and body text
Matched to known operator names
Normalized lab names (Core Labs, Weatherford, etc.)
Geological formation and member identification
Full county name, 2-letter state abbreviation
Oil/gas field identification from document context
Normalized to YYYY-MM-DD from any format
Built for the Worst Documents
Faded 1960s photocopies. Handwritten notes. Rotated pages. Multi-column layouts. If it was important enough to scan, it's important enough for us to read.
Drag & Drop Upload
Upload single files or batch hundreds at once. PDFs, TIFFs, JPEGs, PNGs - even rotated and degraded scans. Processing starts automatically.
Metadata Extraction
API numbers, well names, operators, labs, formations, counties - all parsed from headers, cover pages, and body text. Standardized to PPDM format.
Page Classification
Every page is classified by structure and content type. Core analysis, mud log shows, completion data, DSTs - each gets a specialized extraction pipeline.
Multi-Table Merging
Tables that span multiple pages are automatically detected and merged. Continuation markers, depth range analysis, and header matching ensure nothing is missed.
Excel & CSV Export
Download structured data as Excel spreadsheets or CSV files, ready for import into Petrel, Kingdom, or any analysis tool. One file or batch export entire orders.
QC Portal & Review
Side-by-side comparison of extracted data against the original document. Flag issues, approve extractions, and track quality across your entire dataset.
Upload Anything
NDepth handles any scan format, any orientation, any condition.
Ready to Try NDepth?
Start with $10 free credit. No credit card required. Upload your first document in minutes.