SOTAVerified

Key Information Extraction

Key Information Extraction (KIE) is aimed at extracting structured information (e.g. key-value pairs) from form-style documents (e.g. invoices), which makes an important step towards intelligent document understanding.

Papers

Recently AddedMost HypedMost ActiveNeeds VerificationMost Verified

Showing 1–25 of 74 papers

TitleStatusHype
TextMonkey: An OCR-Free Large Multimodal Model for Understanding DocumentCode5
LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document UnderstandingCode2
OCRBench: On the Hidden Mystery of OCR in Large Multimodal ModelsCode2
A Bounding Box is Worth One Token: Interleaving Layout and Text in a Large Language Model for Document UnderstandingCode2
LayoutLM: Pre-training of Text and Layout for Document Image UnderstandingCode2
Form-NLU: Dataset for the Form Natural Language UnderstandingCode1
Reading Order Matters: Information Extraction from Visually-rich Documents by Token Path PredictionCode1
Exploring OCR Capabilities of GPT-4V(ision) : A Quantitative and In-depth EvaluationCode1
LAMBERT: Layout-Aware (Language) Modeling for information extractionCode1
BROS: A Pre-trained Language Model Focusing on Text and Layout for Better Key Information Extraction from DocumentsCode1
Doc2Graph: a Task Agnostic Document Understanding Framework based on Graph Neural NetworksCode1
DocILE Benchmark for Document Information Localization and ExtractionCode1
PEneo: Unifying Line Extraction, Line Grouping, and Entity Linking for End-to-end Document Pair ExtractionCode1
GenKIE: Robust Generative Multimodal Document Key Information ExtractionCode1
Key Information Extraction From Documents: Evaluation And GeneratorCode1
Modeling Layout Reading Order as Ordering Relations for Visually-rich Document UnderstandingCode1
KVP10k : A Comprehensive Dataset for Key-Value Pair Extraction in Business DocumentsCode1
PICK: Processing Key Information Extraction from Documents using Improved Graph Learning-Convolutional NetworksCode1
ERNIE-Layout: Layout Knowledge Enhanced Pre-training for Visually-rich Document UnderstandingCode1
DoSA : A System to Accelerate Annotations on Business Documents with Human-in-the-LoopCode0
Class-Agnostic Region-of-Interest Matching in Document ImagesCode0
LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document UnderstandingCode0
GraphRevisedIE: Multimodal Information Extraction with Graph-Revised NetworkCode0
GeoLayoutLM: Geometric Pre-training for Visual Information ExtractionCode0
AMuRD: Annotated Arabic-English Receipt Dataset for Key Information Extraction and ClassificationCode0
Show:102550
← PrevPage 1 of 3Next →
All datasetsCORDSROIEKleister NDASIMARA

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1RORE (GeoLayoutLM)F198.52—Unverified
2GeoLayoutLMF197.97—Unverified
3LayoutLMv3 LargeF197.46—Unverified
4LayoutMask (large)F197.19—Unverified
5LayoutMask (base)F196.99—Unverified
6TPP (LayoutMask)F196.92—Unverified
7LILTF196.07—Unverified
8LayoutLMv2LARGEF196.01—Unverified
9LayoutLMv2BASEF194.95—Unverified
#ModelMetricClaimedVerifiedStatus
1LayoutLMv2LARGE (Excluding OCR mismatch)F197.81—Unverified
2RORE (GeoLayoutLM)F196.97—Unverified
3LayoutLMv2LARGEF196.61—Unverified
4LayoutLMv2BASEF196.25—Unverified
5ChatGPT 3.5 SpatialFormatAccuracy77—Unverified
#ModelMetricClaimedVerifiedStatus
1LayoutLMv2LARGEF185.2—Unverified
2LayoutLMv2BASEF183.3—Unverified
3LAMBERT (75M)F180.42—Unverified
#ModelMetricClaimedVerifiedStatus
1DANF1 (%)95.05—Unverified