File size: 1,587 Bytes
6a268d5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2b49742
 
 
 
 
 
6a268d5
 
 
2b49742
902fec2
6a268d5
 
 
 
 
 
 
2b49742
6a268d5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
---
license: mit
task_categories:
  - table-question-answering
  - image-to-text
tags:
  - table-extraction
  - benchmark
  - fintabnet
  - document-ai
  - docld
pretty_name: DocLD FinTabNet Benchmark
size_categories:
  - n<1K
---

# DocLD FinTabNet Benchmark Results

Benchmark results for [DocLD](https://docld.com) table extraction on the
[FinTabNet](https://paperswithcode.com/dataset/fintabnet) dataset.

## Results Summary

| Metric | Value |
|--------|-------|
| **Mean Accuracy** | 82.1% |
| **Median** | 83.2% |
| **P25 / P75** | 73.3% / 97.4% |
| **Min / Max** | 22.7% / 100.0% |
| **Scored Samples** | 500 |
| **Total Samples** | 500 |

## Methodology

- **Dataset**: [FinTabNet_OTSL](https://huggingface.co/datasets/docling-project/FinTabNet_OTSL) — 500 samples from the test split
- **Extraction**: DocLD vision-based table extraction
- **Scoring**: Needleman-Wunsch hierarchical alignment (same as [RD-TableBench](https://github.com/reductoai/rd-tablebench))
- **Output**: HTML tables with rowspan/colspan for merged cells

## Comparison

| Provider | FinTabNet Accuracy |
|----------|-------------------|
| **DocLD** | **82.1%** |
| GTE (IBM) | ~78% |
| TATR (Microsoft) | ~65% |

## Files

- `results.json` — Full benchmark results with per-sample scores
- `predictions/` — HTML predictions for each sample
- `charts/` — Visualization PNGs

## Links

- [DocLD](https://docld.com)
- [Blog Post](https://docld.com/blog/docld-fintabnet)
- [Benchmark Code](https://github.com/Doc-LD/fintabnet-bench)
- [RD-TableBench Results](https://docld.com/blog/docld-tablebench)