Scene Text Recognition2020en

svtp

Dataset from Papers With Code

Metrics:accuracy, cer, wer, f1

Current State of the Art

DTrOCR 105M

Unknown

98.6

accuracy

accuracy Progress Over Time

Showing 5 breakthroughs from Mar 2021 to Aug 2023

Key Milestones

Mar 2021

ABINet-LV

ABINet Language-Vision variant. CVPR 2021.

89.5

Source

Nov 2021

MATRN

From paper: Multi-modal Text Recognition Networks: Interactive Enhancements between Visual and Semantic Features

90.6

+1.2%

Source

Jul 2022

PARSeq

Lowercase alphanum eval. ECCV 2022.

96.9

+7.0%

Source

Sep 2022

MGP-STR

From paper: Multi-Granularity Prediction for Scene Text Recognition

98.3

+1.4%

Source

Aug 2023

DTrOCR 105MCurrent SOTA

From paper: DTrOCR: Decoder-only Transformer for Optical Character Recognition

98.6

+0.3%

Source

Total Improvement

10.2%

Time Span

2y 5m

Breakthroughs

Current SOTA

98.6

Top Models Performance Comparison

Top 10 models ranked by accuracy

Best Score

98.6

Top Model

DTrOCR 105M

Models Compared

Score Range

7.0

accuracyPrimary

#	Model	Score	Paper / Code	Date
1	DTrOCR 105M	98.6	DTrOCR: Decoder-only Transformer for Optical Character Recognition Code	Aug 2023
2	MGP-STR	98.3	Multi-Granularity Prediction for Scene Text Recognition Code	Sep 2022
3	CLIP4STR-L (DataComp-1B)	98.1	CLIP4STR: A Simple Baseline for Scene Text Recognition with Pre-trained Vision-Language Model Code	May 2023
4	CLIP4STR-L	97.4	CLIP4STR: A Simple Baseline for Scene Text Recognition with Pre-trained Vision-Language Model Code	May 2023
5	CLIP4STR-B Research	97.2	CLIP4STR: A Simple Baseline for Scene Text Recognition with Pre-trained Vision-Language Model Code	May 2023
6	PARSeqOpen Source Research	96.9	Scene Text Recognition with Permuted Autoregressive Sequence Models	Jul 2022
7	CPPD	96.7	Context Perception Parallel Decoder for Scene Text Recognition Code	Jul 2023
8	CCD-ViT-Base	96.1	Self-supervised Character-to-Character Distillation for Text Recognition Code	Nov 2022
9	CCD-ViT-Small	92.7	Self-supervised Character-to-Character Distillation for Text Recognition Code	Nov 2022
10	CCD-ViT-Tiny	91.6	Self-supervised Character-to-Character Distillation for Text Recognition Code	Nov 2022
11	S-GTR	90.6	Visual Semantics Allow for Textual Reasoning Better in Scene Text Recognition Code	Dec 2021
12	MATRN Research	90.6	Multi-modal Text Recognition Networks: Interactive Enhancements between Visual and Semantic Features Code	Nov 2021
13	SIGA_T	90.5	Self-supervised Implicit Glyph Attention for Text Recognition Code	Mar 2022
14	CDistNet (Ours)	89.77	CDistNet: Perceiving Multi-Domain Character Distance for Robust Text Recognition Code	Nov 2021
15	ABINet-LVOpen Source Fang et al.	89.5	Read Like Humans: Autonomous, Bidirectional and Iterative Language Modeling for Scene Text Recognition	Mar 2021
16	DiffusionSTR	89.2	DiffusionSTR: Diffusion Model for Scene Text Recognition	Jun 2023
17	DPAN	89	Look Back Again: Dual Parallel Attention Network for Accurate and Robust Scene Text RecognitionCode	Aug 2021
18	TrOCR-large 558M	88.1	TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models	Sep 2021
19	TrOCR-base 334M	86.9	TrOCR: Transformer-based Optical Character Recognition with Pre-trained Models	Sep 2021