Near the end of the day, I was asked to convert invoices and scanned tables into Excel files. The images were slightly blurry, misaligned, and had inconsistent table borders.
The usual approach was slow and frustrating. I had to type data row by row, constantly zoom in and out, double-check numbers, and still worry about missing or incorrect values.
I tried existing image-to-Excel and OCR tools, but ran into the same problems repeatedly. Table structure reconstruction was unreliable, OCR errors were frequent, and batch processing was either limited or unusable.
I then experimented with AI-based OCR models. The recognition accuracy improved, but the workflow was still painful. Uploading images one by one, writing prompts, and copying results manually does not scale when dealing with many invoices.
So I built a web tool focused on one thing only: converting table images into usable Excel files with minimal manual effort.
The current workflow is simple. You upload one or multiple images, the system detects text and table structure, you preview the generated Excel file, and then download it.
I mainly use it for invoices, receipts, printed tables, and screenshots. The output is not perfect in every edge case, but it is significantly better than manual typing and good enough for real-world work.
[https://jpg2excel.app](https://jpg2excel.app)
This is not a must-have tool. But if you regularly deal with table data trapped in images, it may save you a few hours of manual work.