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

Showing 110 of 74 papers

TitleStatusHype
TextMonkey: An OCR-Free Large Multimodal Model for Understanding DocumentCode5
LayoutLM: Pre-training of Text and Layout for Document Image UnderstandingCode2
A Bounding Box is Worth One Token: Interleaving Layout and Text in a Large Language Model for Document UnderstandingCode2
OCRBench: On the Hidden Mystery of OCR in Large Multimodal ModelsCode2
LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document UnderstandingCode2
GenKIE: Robust Generative Multimodal Document Key Information ExtractionCode1
Form-NLU: Dataset for the Form Natural Language UnderstandingCode1
Key Information Extraction From Documents: Evaluation And GeneratorCode1
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
Show:102550
← PrevPage 1 of 8Next →

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1RORE (GeoLayoutLM)F198.52Unverified
2GeoLayoutLMF197.97Unverified
3LayoutLMv3 LargeF197.46Unverified
4LayoutMask (large)F197.19Unverified
5LayoutMask (base)F196.99Unverified
6TPP (LayoutMask)F196.92Unverified
7LILTF196.07Unverified
8LayoutLMv2LARGEF196.01Unverified
9LayoutLMv2BASEF194.95Unverified
#ModelMetricClaimedVerifiedStatus
1LayoutLMv2LARGE (Excluding OCR mismatch)F197.81Unverified
2RORE (GeoLayoutLM)F196.97Unverified
3LayoutLMv2LARGEF196.61Unverified
4LayoutLMv2BASEF196.25Unverified
5ChatGPT 3.5 SpatialFormatAccuracy77Unverified
#ModelMetricClaimedVerifiedStatus
1LayoutLMv2LARGEF185.2Unverified
2LayoutLMv2BASEF183.3Unverified
3LAMBERT (75M)F180.42Unverified
#ModelMetricClaimedVerifiedStatus
1DANF1 (%)95.05Unverified