How to Convert Screenshot to Markdown Table on Windows

You have a screenshot with structured data — an invoice, a config table, error codes. You need it in Markdown for GitHub, Notion, Obsidian, or a README. Manually typing pipes and aligning columns is painful.

CaptureOCR exports Markdown tables directly. Here’s how.

The Problem with Manual Markdown Tables

Manual Approach	Pain Points
Type `\| Header \| Header \|` then each row	Slow, error-prone, alignment breaks
Copy from Excel → Markdown converter	Extra tool, loses coordinates, no confidence
Screenshot → OCR → manual formatting	Double work, easy to misalign columns

CaptureOCR: Screenshot → Markdown in 30 Seconds

Step 1: Capture Regions

Ctrl+Shift+X → drag around each data region. Auto-numbered ①②③.

For a table, you have two options:

Capture whole table as one region → OCR returns text lines, export as Markdown
Capture each cell/column as separate regions → more control, each cell gets its own row

Step 2: Run OCR

Fast mode (tiny model) for clean tables. Accurate mode for messy scans.

Step 3: Export Markdown

Click Export → Markdown. You get:

| # | Region (x,y,w,h) | Line | Confidence | Text |
|---|---|---|---|---|
| 1 | (120,45,380,80) | 1 | 0.998 | Config: database.url=postgres://... |
| 1 | (120,45,380,80) | 2 | 0.995 | Config: database.pool=20 |
| 2 | (120,140,380,60) | 1 | 0.992 | Server: api.example.com:443 |

Every row includes:

Region index (which capture it came from)
Screen coordinates (x,y,width,height)
Line number within region
Confidence score (0.00–1.00)
Recognized text (pipes and newlines escaped)

Markdown Table Features

Feature	How It Works
Pipe escaping	Any `\|` in text becomes `\\|` — table never breaks
Newline handling	Multi-line OCR text becomes `<br>` in cell
Backslash escaping	`\` becomes `\\` — no parsing surprises
Coordinates included	Trace any row back to exact screen location
Confidence scores	Spot low-confidence rows needing review

Paste Directly Into

Platform	Works?
GitHub/GitLab README	✅ Native render
Notion	✅ Paste → converts to table block
Obsidian	✅ Native
VS Code Markdown preview	✅
Confluence (Markdown macro)	✅
Docusaurus/GitBook/Hugo	✅

Real Example: Config File to Docs

Source: A YAML config file on screen

database:
  host: db.prod.internal
  port: 5432
  name: orders
  pool: 20
redis:
  host: cache.internal
  port: 6379

CaptureOCR workflow:

Capture whole config as one region
Fast mode OCR (~78ms)
Export Markdown
Paste into docs → instant formatted table

Result in GitHub: | # | Region (x,y,w,h) | Line | Confidence | Text | |—|—|—|—|—| | 1 | (100,50,400,300) | 1 | 0.998 | database: | | 1 | (100,50,400,300) | 2 | 0.995 | host: db.prod.internal | | 1 | (100,50,400,300) | 3 | 0.992 | port: 5432 |

Pro Tips

For Actual Tables (Grid Data)

If your screenshot is a visual table (rows/columns visible):

Capture the whole table as one region
OCR reads row-by-row
Export Markdown → you get sequential lines
Post-process in Excel: Paste CSV → pivot into grid

CaptureOCR exports line-level rows, not cell-level grid reconstruction. A 4×6 table becomes 24 rows with region coordinates. Many users paste CSV into Excel and use “Text to Columns” or pivot tables.

For Documentation

Add a note in your docs:

Table generated from screenshot via CaptureOCR. Coordinates reference source UI.

Confidence Thresholding

Low-confidence rows (under 0.90) often need review. In your editor, search for 0.8 or 0.7 in the Markdown to find them fast.

Compare: Manual vs CaptureOCR

Task	Manual	CaptureOCR
10-row table	5–10 min	30 sec
Pipe escaping	Manual	Automatic
Coordinates	None	Included
Confidence scores	None	Included
Reproducibility	Low	High (session package)

Get the Workflow

Install CaptureOCR ($2.99 one-time)

Ctrl+Shift+X → capture your data
Export Markdown → paste anywhere

Related reads: