AI Paper Atlas

Research map / Benchmarks / FUNSD

FUNSD leaderboard

11 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on FUNSD. Best F1: 92.86 (GeoLayoutLM, CVPR 2023).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

FUNSD · F1

11 papers, each with its best reported number for its own method; higher is better.

#F1 ↑MethodPaperTopic
192.86GeoLayoutLMGeoLayoutLM: Geometric Pre-Training for Visual Information Extraction
CVPR 2023 · Table 1
Human and embodied modeling
291.7DocMambaDocMamba: Efficient Document Pre-training with State Space Model
AAAI 2025 · Table 1
Attention and transformers
391.62UDOP
T5-large
Unifying Vision, Text, and Layout for Universal Document Processing
CVPR 2023 · Table: Performance on FUNSD, C
Human and embodied modeling
491.55AETNet_base
LayoutLMv3_base
Alignment-Enriched Tuning for Patch-Level Pre-trained Document Image Models
AAAI 2023 · Table: Performance comparison
CLIP and zero-shot
588.89DocFormerv2-largeDocFormerv2: Local Features for Document Understanding
AAAI 2024 · Table: FUNSD comparison
Human and embodied modeling
688.23LDM
SAM-BASE
LDP: Generalizing to Multilingual Visual Information Extraction by Language Decoupled Pretraining
AAAI 2025 · Table, Performance on FUNSD
Human and embodied modeling
787.96UDocUniDoc: Unified Pretraining Framework for Document Understanding
NeurIPS 2021 · Table: Comparison with state-o
Human and embodied modeling
884.67SAIL
GPT-4
SAIL: Sample-Centric In-Context Learning for Document Information Extraction
AAAI 2025 · Quantitative results table
Language, safety and interpretability
982.23KVPFormer
LayoutLM_BASE
A Question-Answering Approach to Key Value Pair Extraction from Form-Like Document Images
AAAI 2023 · Table 1
Visual question answering
1074.79LayoutLMv3 with RIDGE pre-training, 10-shot
LayoutLMv3
Relation-Rich Visual Document Generator for Visual Information Extraction
CVPR 2025 · Table 5
Human and embodied modeling
1 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.