ExtractFoxExtract data from charts and graphs
Drop a screenshot or PDF page of a chart and get the underlying numbers as a clean table. Bar, line, pie, scatter, area — all extracted into rows you can sum, plot, or paste into a spreadsheet.
What should we pull from this chart?
Or pick specific fields
Or describe it yourself
Why this matters
Charts in reports and dashboards bury their data. Manual reconstruction with a ruler is slow and lossy. ExtractFox uses a vision model that reads axis labels, gridlines, and series legends together — so it returns numeric values, not pixel positions. An analyst screenshots a competitor's investor-deck bar chart because the underlying spreadsheet isn't published — reconstructing Q1–Q4 values with a ruler takes 20 minutes and the y-axis scale is logarithmic. Dashboard exports embed charts as images inside PowerPoint slides, so the only copy of last quarter's churn rate is pixels on slide 14.
How it works
- Step 1Upload the chart image or PDF
Screenshot from a dashboard, page from a PDF report, or photo of a printed chart. PNG, JPG, WEBP, HEIC, or PDF.
- Step 2Get series and values
Each series becomes a column or set of rows, with labels read from axes and legend.
- Step 3Export as a table
Download as CSV, JSON, or .xlsx. Rebuild the chart anywhere.
Common use cases
Sample output
Example: bar chart of revenue by region extracted to rows
Request: "extract this bar chart as rows of region and revenue"
Result:
{
"title": "Revenue by region (USD millions)",
"series": "Revenue (USD millions)",
"rows": [
{ "region": "North America", "revenue": 482.1 },
{ "region": "Europe", "revenue": 318.4 },
{ "region": "APAC", "revenue": 246.7 },
{ "region": "LATAM", "revenue": 92.3 }
]
}Frequently asked questions
How do I extract data from a chart in a PDF or image?+
Upload the chart screenshot, PDF page, or photo on this page. ExtractFox reads the axes, legend, and bars/points to return the underlying values as a table you can download as CSV or Excel.
Does it work on bar, line, pie, and scatter charts?+
Yes. Bar/column, line, pie/donut, area, scatter, and stacked variants all work. Heavily annotated or 3D charts are harder but usually still extractable for the main series.
How accurate are the numbers compared to the source?+
Where the chart shows numeric labels (data labels on each bar/point), accuracy is essentially exact. Where values must be read off the axis, expect the result to be within ~1–3% of true values depending on resolution.
Can I extract data from a chart in Excel?+
If the chart is on a screenshot or exported as an image, yes — paste the screenshot here. If you have the .xlsx, the underlying data is already on the worksheet the chart was built from; no extraction needed.
Can it extract data from charts inside a PDF report?+
Yes. Upload the whole PDF and ExtractFox finds and extracts charts page by page. For multi-chart reports, run extraction once per chart for the cleanest results.
What about Gantt charts and org charts?+
Gantt charts extract into tasks with start/end dates and owners. Org charts extract into a parent-child node list. Use the free-text description box to specify exactly the structure you want.
Can it read dual-axis charts with two different y-scales?+
Yes. Each axis is read separately and series are mapped to the correct scale. Specify in the description box if you want left-axis and right-axis values in separate columns.
What about charts where bars lack data labels — only axis ticks?+
Values are estimated from axis position — typically within 1–3% of true values depending on image resolution. Charts with printed data labels on each bar or point are essentially exact.