HyperAI초신경
홈뉴스최신 연구 논문튜토리얼데이터셋백과사전SOTALLM 모델GPU 랭킹컨퍼런스
전체 검색
소개
HyperAI초신경

Handwritten Text Recognition On Iam

평가 지표

CER
WER

평가 결과

이 벤치마크에서 각 모델의 성능 결과

모델 이름
CER
WER
Paper TitleRepository
Decouple Attention Network6.419.6Decoupled Attention Network for Text Recognition
FPHR+Aug Paragraph Level (~145 dpi)6.3-Full Page Handwriting Recognition via Image to Sequence Extraction
Transformer w/ CNN7.62-Pay Attention to What You Read: Non-recurrent Handwritten Text-Line Recognition-
Self-Attention + CTC + language model2.75-Rethinking Text Line Recognition Models-
Start, Follow, Read6.423.2Start, Follow, Read: End-to-End Full-Page Handwriting Recognition
HTR-VT(line-level)4.714.9HTR-VT: Handwritten Text Recognition with Vision Transformer
FPHR Paragraph Level (~145 dpi)6.7-Full Page Handwriting Recognition via Image to Sequence Extraction
TrOCR-small 62M4.22-TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models
Leaky LP Cell6.615.9No Padding Please: Efficient Neural Handwriting Recognition
Transformer + CNN2.96-Rethinking Text Line Recognition Models-
Transformer w/ CNN (+synth)4.67-Pay Attention to What You Read: Non-recurrent Handwritten Text-Line Recognition-
TrOCR-large 558M2.89-TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models
FPHR+Aug Line Level (~145 dpi)6.5-Full Page Handwriting Recognition via Image to Sequence Extraction
VAN4.3216.24End-to-end Handwritten Paragraph Text Recognition Using a Vertical Attention Network
DTrOCR 105M2.38-DTrOCR: Decoder-only Transformer for Optical Character Recognition
Easter2.06.21-Easter2.0: Improving convolutional models for handwritten text recognition
LSTM with attention4.87-Evaluating Sequence-to-Sequence Models for Handwritten Text Recognition
TrOCR-base 334M3.42-TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models
0 of 18 row(s) selected.