Codesota · Benchmark · inverse-textHome/Leaderboards/Vision & Documents/Document OCR/inverse-text
Unknown

inverse-text.

inverse-text is a state-of-the-art machine learning benchmark indexed on Codesota. This page tracks published model results, top scores per metric, and the SOTA timeline for inverse-text.

Paper ↗Leaderboard ↓
§ 01 · SOTA history

Year over year.

§ 02 · Leaderboard

Results by metric.

Found a wrong score or missing run?
Use row edits to send a sourced correction into moderation.
Add / edit result ↗Report issue ↗

F Measure Full Lexicon

F Measure Full Lexicon is the reported evaluation metric for inverse-text. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.

Higher is better

Trust tiers for F Measure Full Lexiconverifiedpapervendorcommunityunverified
RankModelTrustScoreYearLinksFix
01OmniParser
From paper: OmniParser: A Unified Framework for Text Spotting, Key Information Extraction and Table Recognition (CVPR 2024). End-to-end F-measure, Full setting on Inverse-Text. Score sourced from comparison table in LRANet++ (arXiv:2511.05818).
verified88.92024Source ↗Looks wrong?
02DNTextSpotter (ViTAEv2-S)
From paper: DNTextSpotter: Arbitrary-Shaped Scene Text Spotting via Improved Denoising Training (ACM MM 2024). End-to-end F-measure, Full setting, ViTAEv2-S backbone.
verified83.82024Source ↗Looks wrong?
03LRANet++
From paper: LRANet++: Low-Rank Approximation Network for Accurate and Efficient Text Spotting (IEEE TPAMI 2025). End-to-end F-measure, Full setting on Inverse-Text.
verified83.52025Source ↗Looks wrong?
04LSGSpotter
From paper: Arbitrary Reading Order Scene Text Spotter with Local Semantics Guidance (AAAI 2025). End-to-end F-measure, Full setting on Inverse-Text.
verified82.32024Source ↗Looks wrong?
05DNTextSpotter (ResNet-50)
From paper: DNTextSpotter: Arbitrary-Shaped Scene Text Spotting via Improved Denoising Training (ACM MM 2024). End-to-end F-measure, Full setting, ResNet-50 backbone.
verified81.62024Source ↗Looks wrong?
06IAST
From paper: Inverse-like Antagonistic Scene Text Spotting via Reading-Order Estimation and Dynamic Sampling (TIP 2024). End-to-end F-measure, Full lexicon setting on Inverse-Text.
verified80.62024Source ↗Looks wrong?
07SwinTextSpotter v2
From paper: SwinTextSpotter v2: Towards Better Synergy for Scene Text Spotting (IJCV 2025). End-to-end F-measure, Full setting, without TextOCR pre-training.
verified76.52025Source ↗Looks wrong?
08DeepSolo (ViTAEv2-S, TextOCR)
From paper: DeepSolo++: Let Transformer Decoder with Explicit Points Solo for Multilingual Text Spotting
verified75.82023Paper ↗Code ↗Looks wrong?
09SPTS v2
From comparison table in: Arbitrary Reading Order Scene Text Spotter with Local Semantics Guidance (AAAI 2025, arXiv:2412.10159). SPTSv2 end-to-end F-measure, Full setting on Inverse-Text.
verified74.92025Source ↗Looks wrong?
10DeepSolo (ResNet-50, TextOCR)
From paper: DeepSolo++: Let Transformer Decoder with Explicit Points Solo for Multilingual Text Spotting
verified71.22023Paper ↗Code ↗Looks wrong?
11SwinTextSpotter
From paper: SwinTextSpotter: Scene Text Spotting via Better Synergy between Text Detection and Text Recognition
verified67.92022Paper ↗Code ↗Looks wrong?
12DeepSolo (ResNet-50)
From paper: DeepSolo++: Let Transformer Decoder with Explicit Points Solo for Multilingual Text Spotting
verified53.92023Paper ↗Code ↗Looks wrong?
13ABCNet v2
From paper: ABCNet v2: Adaptive Bezier-Curve Network for Real-time End-to-end Text Spotting
verified47.42021Paper ↗Code ↗Looks wrong?
14SPTS
From paper: SPTS: Single-Point Text Spotting
verified46.22021Paper ↗Code ↗Looks wrong?
15MaskTextSpotter v2
From paper: Mask TextSpotter: An End-to-End Trainable Neural Network for Spotting Text with Arbitrary Shapes
verified43.52018Paper ↗Code ↗Looks wrong?
16TESTR
From paper: Text Spotting Transformers
verified41.62022Paper ↗Code ↗Looks wrong?
17ABCNet
From paper: ABCNet: Real-time Scene Text Spotting with Adaptive Bezier-Curve Network
verified34.32020Paper ↗Code ↗Looks wrong?

F Measure No Lexicon

F Measure No Lexicon is the reported evaluation metric for inverse-text. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.

Higher is better

Trust tiers for F Measure No Lexiconverifiedpapervendorcommunityunverified
RankModelTrustScoreYearLinksFix
01OmniParser
From paper: OmniParser: A Unified Framework for Text Spotting, Key Information Extraction and Table Recognition (CVPR 2024). End-to-end F-measure, None setting on Inverse-Text. Score sourced from comparison table in LRANet++ (arXiv:2511.05818).
verified842024Source ↗Looks wrong?
02DNTextSpotter (ViTAEv2-S)
From paper: DNTextSpotter: Arbitrary-Shaped Scene Text Spotting via Improved Denoising Training (ACM MM 2024). End-to-end F-measure, None setting, ViTAEv2-S backbone.
verified78.12024Source ↗Looks wrong?
03DNTextSpotter (ResNet-50)
From paper: DNTextSpotter: Arbitrary-Shaped Scene Text Spotting via Improved Denoising Training (ACM MM 2024). End-to-end F-measure, None setting, ResNet-50 backbone.
verified75.92024Source ↗Looks wrong?
04LRANet++
From paper: LRANet++: Low-Rank Approximation Network for Accurate and Efficient Text Spotting (IEEE TPAMI 2025). End-to-end F-measure, None setting on Inverse-Text.
verified75.32025Source ↗Looks wrong?
05LSGSpotter
From paper: Arbitrary Reading Order Scene Text Spotter with Local Semantics Guidance (AAAI 2025). End-to-end F-measure, None setting on Inverse-Text.
verified73.72024Source ↗Looks wrong?
06DeepSolo (ViTAEv2-S, TextOCR)
From paper: DeepSolo++: Let Transformer Decoder with Explicit Points Solo for Multilingual Text Spotting
verified68.82023Paper ↗Code ↗Looks wrong?
07IAST
From paper: Inverse-like Antagonistic Scene Text Spotting via Reading-Order Estimation and Dynamic Sampling (TIP 2024). End-to-end F-measure, None (lexicon-free) setting on Inverse-Text.
verified68.82024Source ↗Looks wrong?
08SwinTextSpotter v2
From paper: SwinTextSpotter v2: Towards Better Synergy for Scene Text Spotting (IJCV 2025). End-to-end F-measure, None setting, without TextOCR pre-training.
verified64.82025Source ↗Looks wrong?
09DeepSolo (ResNet-50, TextOCR)
From paper: DeepSolo++: Let Transformer Decoder with Explicit Points Solo for Multilingual Text Spotting
verified64.62023Paper ↗Code ↗Looks wrong?
10SPTS v2
From comparison table in: Arbitrary Reading Order Scene Text Spotter with Local Semantics Guidance (AAAI 2025, arXiv:2412.10159). SPTSv2 end-to-end F-measure, None setting on Inverse-Text.
verified63.42025Source ↗Looks wrong?
11SwinTextSpotter
From paper: SwinTextSpotter: Scene Text Spotting via Better Synergy between Text Detection and Text Recognition
verified55.42022Paper ↗Code ↗Looks wrong?
12DeepSolo (ResNet-50)
From paper: DeepSolo++: Let Transformer Decoder with Explicit Points Solo for Multilingual Text Spotting
verified48.52023Paper ↗Code ↗Looks wrong?
13MaskTextSpotter v2
From paper: Mask TextSpotter: An End-to-End Trainable Neural Network for Spotting Text with Arbitrary Shapes
verified392018Paper ↗Code ↗Looks wrong?
14SPTS
From paper: SPTS: Single-Point Text Spotting
verified38.32021Paper ↗Code ↗Looks wrong?
15ABCNet v2
From paper: ABCNet v2: Adaptive Bezier-Curve Network for Real-time End-to-end Text Spotting
verified34.52021Paper ↗Code ↗Looks wrong?
16TESTR
From paper: Text Spotting Transformers
verified34.22022Paper ↗Code ↗Looks wrong?
17ABCNet
From paper: ABCNet: Real-time Scene Text Spotting with Adaptive Bezier-Curve Network
verified22.22020Paper ↗Code ↗Looks wrong?
§ 04 · Submit a result

Add to the leaderboard.

← Back to Document OCR