Codesota · Computer Vision · Scene Text Detection · Total-TextTasks/Computer Vision/Scene Text Detection
Scene Text Detection · benchmark dataset · 2017 · EN

Total-Text.

Curved text benchmark. 1555 images with polygon annotations.

Paper ↗Submit a result ↵
§ 01 · Leaderboard

Best published scores.

126 results indexed across 6 metrics. Shaded row marks current SOTA; ties broken by submission date.

Primary
f1 · higher is better
All metrics
f-measure, f-measure-full-lexicon, f-measure-no-lexicon, fps, precision, recall
f-measure
33 rows
#ModelOrgSubmittedPaper / codef-measure
01DAT-SEGOSSWan et al. (Baidu)May 2024Towards Unified Multi-granularity Text Detection with In…92.01
02DAT-DETOSSWan et al. (Baidu)May 2024Towards Unified Multi-granularity Text Detection with In…90.98
03MixNet—Aug 2023MixNet: Toward Accurate Detection of Challenging Scene T… · code90.50
04SRFormer (ResNet-50)—Aug 2023SRFormer: Text Detection Transformer with Incorporated S… · code90
05ERRNet (with pre-training)OSSSu et al.Dec 2024Explicit Relational Reasoning Network for Scene Text Det…89.90
06TextMambaOSSZhao et al.Dec 2024TextMamba: Scene Text Detector with Mamba89.20
07LRANetOSSAAAI 2024Jun 2023LRANet: Towards Accurate and Efficient Scene Text Detect… · code89
08DPText-DETR (ResNet-50)—Jul 2022DPText-DETR: Towards Better Scene Text Detection with Dy… · code89
09ERRNetOSSSu et al.Dec 2024Explicit Relational Reasoning Network for Scene Text Det…88.10
10FAST-B-800—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code87.50
11TextFuseNet (ResNeXt-101)—May 2020papers-with-code · code87.50
12I3CL + SSL(ResNet-50)—Aug 2021I3CL:Intra- and Inter-Instance Collaborative Learning fo… · code86.90
13CharNet H-88 (multi-scale)—Oct 2019Convolutional Character Networks · code86.50
14FAST-B-640—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code86.40
15DBNet++ (ResNet-50) (800)—Feb 2022Real-Time Scene Text Detection with Differentiable Binar… · code86
16FAST-B-512—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code85.80
17SA-Text—Nov 2019A method for detecting text of arbitrary shapes in natur…85.60
18CharNet H-88—Oct 2019Convolutional Character Networks · code85.60
19PAN-640—Aug 2019Efficient and Accurate Arbitrary-Shaped Text Detection w… · code85
20FAST-S-512—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code84.90
21DB-ResNet-50 (800)—Nov 2019Real-time Scene Text Detection with Differentiable Binar… · code84.70
22TextCohesion—Apr 2019TextCohesion: Detecting Text for Arbitrary Shapes84.60
23CRAFT—Apr 2019Character Region Awareness for Text Detection · code83.60
24DBNet++ (ResNet-18) (800)—Feb 2022Real-Time Scene Text Detection with Differentiable Binar… · code83.30
25SPCNET—Nov 2018Scene Text Detection with Supervised Pyramid Context Net… · code82.90
26FAST-T-448—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code81.60
27FTSN—Sep 2017Fused Text Segmentation Networks for Multi-oriented Scen…81.30
28TextFiled—Dec 2018TextField: Learning A Deep Direction Field for Irregular… · code80.60
29PSENet-4s—Mar 2019Shape Robust Text Detection with Progressive Scale Expan… · code79.60
30TextSnake—Jul 2018TextSnake: A Flexible Representation for Detecting Text … · code78.40
31Mask TextSpotter—Jul 2018Mask TextSpotter: An End-to-End Trainable Neural Network… · code61.30
32EAST—Apr 2017EAST: An Efficient and Accurate Scene Text Detector · code42
33Ch,ng et al.—Oct 2017Total-Text: A Comprehensive Dataset for Scene Text Detec… · code36
f-measure-full-lexicon
12 rows
#ModelOrgSubmittedPaper / codef-measure-full-lexicon
01DeepSolo (ViTAEv2-S, TextOCR)—Nov 2022DeepSolo: Let Transformer Decoder with Explicit Points S… · code89.60
02DeepSolo (ResNet-50, TextOCR)—Nov 2022DeepSolo: Let Transformer Decoder with Explicit Points S… · code88.70
03DeepSolo (ResNet-50)—Nov 2022DeepSolo: Let Transformer Decoder with Explicit Points S… · code87
04UNITS—Apr 2023Towards Unified Scene Text Spotting based on Sequence Ge… · code86
05A3S—Feb 2023A3S: Adversarial learning of semantic representations fo…85.10
06SwinTextSpotter—Mar 2022SwinTextSpotter: Scene Text Spotting via Better Synergy … · code84.10
07TESTR—Apr 2022Text Spotting Transformers · code83.90
08MANGO—Dec 2020MANGO: A Mask Attention Guided One-Stage Scene Text Spot… · code83.60
09DEER—Mar 2022DEER: Detection-agnostic End-to-End Recognizer for Scene…83.30
10GLASS—Aug 2022GLASS: Global to Local Attention for Scene-Text Spotting · code83
11MaskTextSpotter v3—Jul 2020Mask TextSpotter v3: Segmentation Proposal Network for R… · code78.40
12ABCNet v2—May 2021ABCNet v2: Adaptive Bezier-Curve Network for Real-time E… · code78.10
f-measure-no-lexicon
12 rows
#ModelOrgSubmittedPaper / codef-measure-no-lexicon
01DeepSolo (ViTAEv2-S, TextOCR)—Nov 2022DeepSolo: Let Transformer Decoder with Explicit Points S… · code83.60
02DeepSolo (ResNet-50, TextOCR)—Nov 2022DeepSolo: Let Transformer Decoder with Explicit Points S… · code82.50
03DeepSolo (ResNet-50)—Nov 2022DeepSolo: Let Transformer Decoder with Explicit Points S… · code79.70
04A3S—Feb 2023A3S: Adversarial learning of semantic representations fo…79.40
05UNITS—Apr 2023Towards Unified Scene Text Spotting based on Sequence Ge… · code78.70
06GLASS—Aug 2022GLASS: Global to Local Attention for Scene-Text Spotting · code76.60
07DEER—Mar 2022DEER: Detection-agnostic End-to-End Recognizer for Scene…74.80
08SwinTextSpotter—Mar 2022SwinTextSpotter: Scene Text Spotting via Better Synergy … · code74.30
09TESTR—Apr 2022Text Spotting Transformers · code73.30
10MANGO—Dec 2020MANGO: A Mask Attention Guided One-Stage Scene Text Spot… · code72.90
11MaskTextSpotter v3—Jul 2020Mask TextSpotter v3: Segmentation Proposal Network for R… · code71.20
12ABCNet v2—May 2021ABCNet v2: Adaptive Bezier-Curve Network for Real-time E… · code70.40
fps
9 rows
#ModelOrgSubmittedPaper / codefps
01FAST-T-448—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code152.80
02FAST-S-512—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code115.50
03FAST-B-512—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code93.20
04FAST-B-640—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code67.50
05DBNet++ (ResNet-18) (800)—Feb 2022Real-Time Scene Text Detection with Differentiable Binar… · code48
06FAST-B-800—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code46
07DBNet++ (ResNet-50) (800)—Feb 2022Real-Time Scene Text Detection with Differentiable Binar… · code28
08DPText-DETR (ResNet-50)—Jul 2022DPText-DETR: Towards Better Scene Text Detection with Dy… · code17
09MixNet—Aug 2023MixNet: Toward Accurate Detection of Challenging Scene T… · code15.20
precision
30 rows
#ModelOrgSubmittedPaper / codeprecision
01DAT-SEGOSSWan et al. (Baidu)May 2024Towards Unified Multi-granularity Text Detection with In…95.04
02DAT-DETOSSWan et al. (Baidu)May 2024Towards Unified Multi-granularity Text Detection with In…93.98
03MixNet—Aug 2023MixNet: Toward Accurate Detection of Challenging Scene T… · code93
04ERRNet (with pre-training)OSSSu et al.Dec 2024Explicit Relational Reasoning Network for Scene Text Det…92.60
05SRFormer (ResNet-50)—Aug 2023SRFormer: Text Detection Transformer with Incorporated S… · code92.20
06DPText-DETR (ResNet-50)—Jul 2022DPText-DETR: Towards Better Scene Text Detection with Dy… · code91.80
07LRANetOSSAAAI 2024Jun 2023LRANet: Towards Accurate and Efficient Scene Text Detect… · code90.30
08ERRNetOSSSu et al.Dec 2024Explicit Relational Reasoning Network for Scene Text Det…90.10
09FAST-B-800—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code90
10CharNet H-88—Oct 2019Convolutional Character Networks · code89.90
11FAST-B-640—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code89.90
12I3CL + SSL(ResNet-50)—Aug 2021I3CL:Intra- and Inter-Instance Collaborative Learning fo… · code89.80
13FAST-B-512—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code89.60
14TextMambaOSSZhao et al.Dec 2024TextMamba: Scene Text Detector with Mamba89.50
15PAN-640—Aug 2019Efficient and Accurate Arbitrary-Shaped Text Detection w… · code89.30
16TextFuseNet (ResNeXt-101)—May 2020papers-with-code · code89.20
17DBNet++ (ResNet-50) (800)—Feb 2022Real-Time Scene Text Detection with Differentiable Binar… · code88.90
18FAST-S-512—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code88.30
19CharNet H-88 (multi-scale)—Oct 2019Convolutional Character Networks · code88
20CRAFT—Apr 2019Character Region Awareness for Text Detection · code87.60
21DBNet++ (ResNet-18) (800)—Feb 2022Real-Time Scene Text Detection with Differentiable Binar… · code87.40
22FAST-T-448—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code86.50
23FTSN—Sep 2017Fused Text Segmentation Networks for Multi-oriented Scen…84.70
24PSENet-4s—Mar 2019Shape Robust Text Detection with Progressive Scale Expan… · code84.50
25SPCNET—Nov 2018Scene Text Detection with Supervised Pyramid Context Net… · code83
26TextSnake—Jul 2018TextSnake: A Flexible Representation for Detecting Text … · code82.70
27TextFiled—Dec 2018TextField: Learning A Deep Direction Field for Irregular… · code81.20
28Mask TextSpotter—Jul 2018Mask TextSpotter: An End-to-End Trainable Neural Network… · code69
29EAST—Apr 2017EAST: An Efficient and Accurate Scene Text Detector · code50
30Ch,ng et al.—Oct 2017Total-Text: A Comprehensive Dataset for Scene Text Detec… · code40
recall
30 rows
#ModelOrgSubmittedPaper / coderecall
01DAT-SEGOSSWan et al. (Baidu)May 2024Towards Unified Multi-granularity Text Detection with In…89.16
02TextMambaOSSZhao et al.Dec 2024TextMamba: Scene Text Detector with Mamba88.80
03DAT-DETOSSWan et al. (Baidu)May 2024Towards Unified Multi-granularity Text Detection with In…88.17
04MixNet—Aug 2023MixNet: Toward Accurate Detection of Challenging Scene T… · code88.10
05SRFormer (ResNet-50)—Aug 2023SRFormer: Text Detection Transformer with Incorporated S… · code87.90
06LRANetOSSAAAI 2024Jun 2023LRANet: Towards Accurate and Efficient Scene Text Detect… · code87.80
07ERRNet (with pre-training)OSSSu et al.Dec 2024Explicit Relational Reasoning Network for Scene Text Det…87.30
08DPText-DETR (ResNet-50)—Jul 2022DPText-DETR: Towards Better Scene Text Detection with Dy… · code86.40
09ERRNetOSSSu et al.Dec 2024Explicit Relational Reasoning Network for Scene Text Det…86.10
10TextFuseNet (ResNeXt-101)—May 2020papers-with-code · code85.80
11FAST-B-800—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code85.20
12CharNet H-88 (multi-scale)—Oct 2019Convolutional Character Networks · code85
13I3CL + SSL(ResNet-50)—Aug 2021I3CL:Intra- and Inter-Instance Collaborative Learning fo… · code84.20
14FAST-B-640—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code83.20
15DBNet++ (ResNet-50) (800)—Feb 2022Real-Time Scene Text Detection with Differentiable Binar… · code83.20
16SPCNET—Nov 2018Scene Text Detection with Supervised Pyramid Context Net… · code82.80
17FAST-B-512—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code82.40
18FAST-S-512—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code81.70
19CharNet H-88—Oct 2019Convolutional Character Networks · code81.70
20PAN-640—Aug 2019Efficient and Accurate Arbitrary-Shaped Text Detection w… · code81
21TextFiled—Dec 2018TextField: Learning A Deep Direction Field for Irregular… · code79.90
22CRAFT—Apr 2019Character Region Awareness for Text Detection · code79.90
23DBNet++ (ResNet-18) (800)—Feb 2022Real-Time Scene Text Detection with Differentiable Binar… · code79.60
24FTSN—Sep 2017Fused Text Segmentation Networks for Multi-oriented Scen…78
25FAST-T-448—Nov 2021FAST: Faster Arbitrarily-Shaped Text Detector with Minim… · code77.20
26PSENet-4s—Mar 2019Shape Robust Text Detection with Progressive Scale Expan… · code75.20
27TextSnake—Jul 2018TextSnake: A Flexible Representation for Detecting Text … · code74.50
28Mask TextSpotter—Jul 2018Mask TextSpotter: An End-to-End Trainable Neural Network… · code55
29EAST—Apr 2017EAST: An Efficient and Accurate Scene Text Detector · code36.20
30Ch,ng et al.—Oct 2017Total-Text: A Comprehensive Dataset for Scene Text Detec… · code33
Fig 2 · Rows sorted by score within each metric. Shaded row marks SOTA. Dates reflect model or paper release where available, otherwise the date Codesota accessed the source.
§ 04 · Literature

34 papers
tied to this benchmark.

Every paper below corresponds to at least one row in the leaderboard above. Click through for the arXiv preprint and, when available, the reference implementation.

§ 06 · Contribute

Have a score that beats
this table?

Submit a checkpoint and a reproduction script. We will run it, publish the score, and — if it takes the top — annotate the step on the progress chart with your name.

Submit a result ↵Read submission guide
What a submission needs
  • 01A public checkpoint or API endpoint
  • 02A reproduction script with frozen commit + seed
  • 03Declared evaluation environment (Python, deps)
  • 04One row per metric declared by this dataset
  • 05A contact so we can follow up on discrepancies
Total-Text — Scene Text Detection benchmark · Codesota | CodeSOTA