Sample RankingRecallPrecisionHmean
GraphDoc100.00%100.00%100.00%
Tencent Youtu75.00%100.00%85.71%
Applica.ai Lambert 2.0 + Excluding OCR Mismatch75.00%100.00%85.71%
Applica.ai Lambert 2.0 + Excluding OCR Errors + Fixing total entity75.00%100.00%85.71%
Applica.ai TILT + Excluding OCR Errors + Fixing total entity75.00%100.00%85.71%
Qubitrics75.00%100.00%85.71%
IE75.00%100.00%85.71%
IE + GCN+ OCR75.00%100.00%85.71%
Linklogis_BeeAI75.00%100.00%85.71%
Character-Aware CNN + Highway + BiLSTM 1.075.00%100.00%85.71%
LAKE75.00%100.00%85.71%
Character-Aware CNN + Highway + BiLSTM 2.075.00%100.00%85.71%
GraphRevisedIE75.00%100.00%85.71%
Enetity detection75.00%75.00%75.00%
BERT with Multi-task Confidence Prediction75.00%75.00%75.00%
H&H Lab75.00%75.00%75.00%
Ping An Property & Casualty Insurance Company75.00%75.00%75.00%
PATECH_CHENGDU_OCR75.00%75.00%75.00%
CITlab Argus Information Extraction (positional & line features, enhanced gt)75.00%75.00%75.00%
BOE_IOT_AIBD75.00%75.00%75.00%
PATECH_CHENGDU_OCR_V275.00%75.00%75.00%
PATech_AICenter75.00%75.00%75.00%
Applica.ai roberta-base-2D75.00%75.00%75.00%
PICK-PAPCIC & XZMU75.00%75.00%75.00%
LayoutLM75.00%75.00%75.00%
LayoutLM + Excluding OCR Mismatch75.00%75.00%75.00%
BERT-MRC75.00%75.00%75.00%
Graph Neural Net with Bert Embeddings75.00%75.00%75.00%
Taikang Insurance Group Research Institute75.00%75.00%75.00%
RoBERTa-base finetuned on business documents75.00%75.00%75.00%
StrucTexT75.00%75.00%75.00%
1129-clm75.00%75.00%75.00%
Multimodal Transformer for Information Extraction50.00%100.00%66.67%
LayoutLM 2.0 (single model)50.00%66.67%57.14%
IE method50.00%66.67%57.14%
1208-cblm50.00%66.67%57.14%
1208-cblm50.00%66.67%57.14%
CHL50.00%66.67%57.14%
Bi-directional LSTM-CNNs-CRF (version2)50.00%50.00%50.00%
HeReceipt-withoutRM50.00%50.00%50.00%
IFLYTEK-textNLP_v250.00%50.00%50.00%
GraphLayout50.00%50.00%50.00%
Location-aware BERT model for Text Information Extraction50.00%50.00%50.00%
HIK_OCR_Exclude_ocr_mismatch50.00%50.00%50.00%
great50.00%50.00%50.00%
RoBERTa-base50.00%50.00%50.00%
N50.00%50.00%50.00%
test250.00%50.00%50.00%
test excluding ocr error50.00%50.00%50.00%
LSTM without OCRerror and RM50.00%50.00%50.00%
LayoutLMv250.00%50.00%50.00%
GIE25.00%50.00%33.33%
layoutlm V225.00%33.33%28.57%
A Simple Method for Key Information Extraction as Character-wise Classification with LSTM25.00%25.00%25.00%
NER with spaCy model25.00%25.00%25.00%
CLOVA OCR25.00%25.00%25.00%
With receipt framing25.00%25.00%25.00%
SituTech_OCR25.00%25.00%25.00%
ASTRI-CCT-MSA25.00%25.00%25.00%
Character & Word BiLSTM Encoder25.00%25.00%25.00%
test25.00%25.00%25.00%
SROIE LSTM - Axel Alejandro Ramos GarcĂ­a25.00%25.00%25.00%
Cool Method Remix25.00%25.00%25.00%
End-to-end learning with PGN25.00%25.00%25.00%
Custom Named Entity Recognition25.00%25.00%25.00%
coldog25.00%25.00%25.00%
lstm deep25.00%25.00%25.00%
Character Level BiLSTM25.00%25.00%25.00%
Token level multi modal bilstm25.00%25.00%25.00%
GEM AI -OCR TEAM25.00%25.00%25.00%
layoutLMtest25.00%25.00%25.00%
PIC-proj25.00%25.00%25.00%
Pytesseract + Character Level LSTM + Regex for Dates25.00%25.00%25.00%
0 shots25.00%25.00%25.00%
5shout25.00%25.00%25.00%
Key Information Extraction from Scanned Receipts0.00%0.00%0.00%
lstm standard method trained 100 epochs constant learning rate0.00%0.00%0.00%
Token level bert embed + bilstm0.00%0.00%0.00%
2shot_tem020.00%0.00%0.00%
1shot_position0.00%0.00%0.00%