You have a chart — in a PDF paper, a slide deck, a report, or a photo — and you need the numbers behind it in Excel. Retyping estimated values off the axes is slow and error-prone. Here is the fast way.

Step-by-step: graph image to Excel

  1. Screenshot or save the graph. Any PNG or JPG works: a figure exported from a paper, a snip from a PDF, even a phone photo of a printed page. Higher resolution gives better precision.
  2. Upload it to Plot2CSV. The plotting area is detected automatically; adjust the crop box if you want to exclude legends or captions.
  3. Calibrate the axes with three taps. Click the origin, the maximum X, and the maximum Y, then type the axis values. Both linear and logarithmic scales are supported — common in scientific figures.
  4. Extract automatically. Every visible line or scatter series is detected and labeled Y1, Y2, Y3… You'll see the traced points overlaid on your graph so you can verify them against the original curves.
  5. Fix anything, then download Excel. Drag points, add missed ones, or type exact values into the table. Click Excel and you get a real .xlsx file with aligned X, Y1, Y2… columns — not a picture of a table, actual numbers.

The whole round trip typically takes one to two minutes.

Why not use AI chat tools to read the chart?

You can paste a chart into an AI chatbot and ask for a table — but you get unverifiable numbers. There is no overlay showing where on the curve each value came from, hallucinated values are common on dense charts, and precision suffers on log scales. A calibration-based extractor is deterministic: every exported number corresponds to a pixel you can see and adjust.

Tips for accurate results

  • Crop tightly to the plotting area — exclude titles and legends so they can't be mistaken for data.
  • Calibrate carefully. Precision depends more on your three calibration points than anything else. Zoom in before tapping.
  • Use log mode for log axes. Reading a log-scale chart as linear is the most common source of wildly wrong data.
  • Review before export. The editable overlay exists exactly for the last 2% of stray points.

Works for CSV and JSON too

Excel is one click; the same extraction also exports CSV (for R, Python, MATLAB pipelines) and JSON (for web apps). Read about when CSV beats Excel, or just start extracting now — it's free and requires no account.