Skip to content

Commit 0ee3cd7

Browse files
docs: refresh benchmark table with current numbers (#137)
Re-ran opendataloader-bench on current main. pdf-inspector improved across the board since the last table: overall 0.78→0.83, tables 0.59→0.66, headings 0.57→0.74 (competitor rows unchanged). Updated the prose — heading detection no longer lags opendataloader, and overall is now within 0.01 of it at ~2.5× the speed. Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
1 parent f5be401 commit 0ee3cd7

1 file changed

Lines changed: 4 additions & 4 deletions

File tree

README.md

Lines changed: 4 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -26,16 +26,16 @@ Evaluated on the [opendataloader-bench](https://github.com/opendataloader-projec
2626

2727
| Engine | Overall | Reading Order (NID) | Tables (TEDS) | Headings (MHS) | Speed (200 docs) |
2828
|---|---|---|---|---|---|
29-
| pdf-inspector | 0.78 | 0.87 | 0.59 | 0.57 | 4s |
29+
| pdf-inspector | 0.83 | 0.88 | 0.66 | 0.74 | 4s |
3030
| opendataloader | 0.84 | 0.91 | 0.49 | 0.74 | 11s |
3131
| pymupdf4llm | 0.73 | 0.89 | 0.40 | 0.41 | 18s |
3232
| markitdown | 0.58 | 0.88 | 0.00 | 0.00 | 8s |
3333

34-
For context, engines that use OCR/ML (docling, marker, mineru) score 0.83-0.88 overall but take 2-180 minutes on the same corpus.
34+
For context, engines that use OCR/ML (docling, marker, mineru) score 0.83-0.88 overall but take 2-180 minutes on the same corpus — pdf-inspector reaches the low end of that range without any OCR, in 4 seconds.
3535

36-
**Where we do well:** Speed (fastest of all engines), reading order, table detection vs other direct-text tools.
36+
**Where we do well:** Speed (fastest of all engines), the best table detection of any engine shown, and heading detection now on par with opendataloader. Overall lands within 0.01 of opendataloader at roughly 2.5× the speed.
3737

38-
**Where we lag:** Heading detection trails opendataloader — many PDFs use bold text at body font size for headings, or headings that are only slightly larger than body text. Table detection trails OCR-based engines that can see visual table structure.
38+
**Where we lag:** Reading order still trails opendataloader slightly, and table structure trails OCR-based engines that can see visual layout.
3939

4040
## Quick start
4141

0 commit comments

Comments
 (0)