Scene Text Detection2020en

ic19-art

Dataset from Papers With Code

Metrics:accuracy, cer, wer, f1
Current State of the Art

CLIP4STR-L (DataComp-1B)

Unknown

86.4

accuracy

accuracy Progress Over Time

Showing 2 breakthroughs from Sep 2022 to May 2023

85.485.786.086.286.5Sep 2022May 2023accuracyDate

Key Milestones

Sep 2022
MGP-STR

From paper: Multi-Granularity Prediction for Scene Text Recognition

85.5
May 2023
CLIP4STR-L (DataComp-1B)Current SOTA

From paper: CLIP4STR: A Simple Baseline for Scene Text Recognition with Pre-trained Vision-Language Model

86.4
+1.1%
Total Improvement
1.1%
Time Span
8m
Breakthroughs
2
Current SOTA
86.4

Top Models Performance Comparison

Top 4 models ranked by accuracy

accuracy1CLIP4STR-L (DataComp-1B)86.4100.0%2CLIP4STR-L85.999.4%3CLIP4STR-B*85.899.3%4MGP-STR85.599.0%0%25%50%75%100%% of best
Best Score
86.4
Top Model
CLIP4STR-L (DataC...
Models Compared
4
Score Range
0.900

accuracyPrimary

h-mean

#ModelScorePaper / CodeDate
1
MixNet
79.7Aug 2023
2
SRFormer (ResNet-50)
79.3Aug 2023
3
TextFuseNet (ResNeXt-101)
78.6
TextFuseNet: Scene Text Detection with Richer Fused FeaturesCode
May 2020
4
DPText-DETR (ResNet-50)
78.1Jul 2022

Related Papers5

Other Scene Text Detection Datasets