Codesota · Benchmark · scut-ctw1500Home/Leaderboards/Vision & Documents/Document OCR/scut-ctw1500
Unknown

scut-ctw1500.

scut-ctw1500 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 scut-ctw1500.

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 ↗

Fps

Fps is the reported evaluation metric for scut-ctw1500. 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 Fpsverifiedpapervendorcommunityunverified
RankModelTrustScoreYearLinksFix
01FAST-T-512
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified129.12021Paper ↗Code ↗Looks wrong?
02FAST-S-512
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified112.92021Paper ↗Code ↗Looks wrong?
03FAST-B-512
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified92.62021Paper ↗Code ↗Looks wrong?
04FAST-B-640
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified66.52021Paper ↗Code ↗Looks wrong?
05PAN
From paper: Mask R-CNN with Pyramid Attention Network for Scene Text Detection
verified65.22018Paper ↗Looks wrong?
06MixNet
From paper: MixNet: Toward Accurate Detection of Challenging Scene Text in the Wild
verified15.22023Paper ↗Code ↗Looks wrong?

Precision

Precision is the reported evaluation metric for scut-ctw1500. 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 Precisionverifiedpapervendorcommunityunverified
RankModelTrustScoreYearLinksFix
01DeepSolo (with pre-training)
Pre-trained on Synth150K+MLT17+IC13+IC15. Source: Table 7, arxiv:2305.19957
verified92.52022Paper ↗Code ↗Looks wrong?
02DPText-DETR (ResNet50)
From paper: DPText-DETR: Towards Better Scene Text Detection with Dynamic Points in Transformer
verified91.72022Paper ↗Code ↗Looks wrong?
03SRFormer (ResNet-50)
From paper: SRFormer: Text Detection Transformer with Incorporated Segmentation and Regression
verified91.62023Paper ↗Code ↗Looks wrong?
04MixNet
From paper: MixNet: Toward Accurate Detection of Challenging Scene Text in the Wild
verified91.42023Paper ↗Code ↗Looks wrong?
05TextMamba
ResNet-50 backbone. Source: Table I, arxiv:2512.06657
verified912024Paper ↗Looks wrong?
06TextFuseNet (ResNeXt-101)
From paper: TextFuseNet: Scene Text Detection with Richer Fused Features
verified89.72020Paper ↗Code ↗Looks wrong?
07I3CL + SSL
From paper: I3CL:Intra- and Inter-Instance Collaborative Learning for Arbitrary-shaped Scene Text Detection
verified88.42021Paper ↗Code ↗Looks wrong?
08EK-Net
ResNet-18 backbone. Source: Table 3, arxiv:2401.11704
verified87.852024Paper ↗Looks wrong?
09FAST-B-640
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified87.82021Paper ↗Code ↗Looks wrong?
10PAN
From paper: Mask R-CNN with Pyramid Attention Network for Scene Text Detection
verified86.82018Paper ↗Looks wrong?
11PAN-640
From paper: Efficient and Accurate Arbitrary-Shaped Text Detection with Pixel Aggregation Network
verified86.42019Paper ↗Code ↗Looks wrong?
12CRAFT
From paper: Character Region Awareness for Text Detection
verified862019Paper ↗Code ↗Looks wrong?
13FAST-B-512
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified85.72021Paper ↗Code ↗Looks wrong?
14FAST-S-512
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified85.62021Paper ↗Code ↗Looks wrong?
15FAST-T-512
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified85.52021Paper ↗Code ↗Looks wrong?
16PSENet-1s
From paper: Shape Robust Text Detection with Progressive Scale Expansion Network
verified82.52018Paper ↗Code ↗Source ↗Looks wrong?
17SLPR
From paper: Sliding Line Point Regression for Shape Robust Scene Text Detection
verified80.12018Paper ↗Code ↗Looks wrong?
18TextSnake
From paper: TextSnake: A Flexible Representation for Detecting Text of Arbitrary Shapes
verified67.92018Paper ↗Code ↗Looks wrong?

F Measure

F Measure is the reported evaluation metric for scut-ctw1500. 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 Measureverifiedpapervendorcommunityunverified
RankModelTrustScoreYearLinksFix
01MixNet
From paper: MixNet: Toward Accurate Detection of Challenging Scene Text in the Wild
verified89.82023Paper ↗Code ↗Looks wrong?
02TextMamba
ResNet-50 backbone. Source: Table I, arxiv:2512.06657
verified89.72024Paper ↗Looks wrong?
03SRFormer (ResNet-50)
From paper: SRFormer: Text Detection Transformer with Incorporated Segmentation and Regression
verified89.62023Paper ↗Code ↗Looks wrong?
04DeepSolo (with pre-training)
Pre-trained on Synth150K+MLT17+IC13+IC15. P=92.5, R=86.3. Source: Table 7, arxiv:2305.19957
verified89.32022Paper ↗Code ↗Looks wrong?
05DPText-DETR (ResNet50)
From paper: DPText-DETR: Towards Better Scene Text Detection with Dynamic Points in Transformer
verified88.82022Paper ↗Code ↗Looks wrong?
06TextFuseNet (ResNeXt-101)
From paper: TextFuseNet: Scene Text Detection with Richer Fused Features
verified87.42020Paper ↗Code ↗Looks wrong?
07I3CL + SSL
From paper: I3CL:Intra- and Inter-Instance Collaborative Learning for Arbitrary-shaped Scene Text Detection
verified86.52021Paper ↗Code ↗Looks wrong?
08EK-Net
ResNet-18 backbone, 40.13 FPS. Source: Table 3, arxiv:2401.11704
verified85.752024Paper ↗Looks wrong?
09PAN
From paper: Mask R-CNN with Pyramid Attention Network for Scene Text Detection
verified852018Paper ↗Looks wrong?
10FAST-B-640
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified84.22021Paper ↗Code ↗Looks wrong?
11PAN-640
From paper: Efficient and Accurate Arbitrary-Shaped Text Detection with Pixel Aggregation Network
verified83.72019Paper ↗Code ↗Looks wrong?
12CRAFT
From paper: Character Region Awareness for Text Detection
verified83.52019Paper ↗Code ↗Looks wrong?
13DB-ResNet50 (1024)
From paper: Real-time Scene Text Detection with Differentiable Binarization
verified83.42019Paper ↗Code ↗Looks wrong?
14FAST-B-512
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified82.92021Paper ↗Code ↗Looks wrong?
15FAST-S-512
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified822021Paper ↗Code ↗Looks wrong?
16FAST-T-512
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified81.52021Paper ↗Code ↗Looks wrong?
17PSENet-1s
From paper: Shape Robust Text Detection with Progressive Scale Expansion Network
verified81.172018Paper ↗Code ↗Source ↗Looks wrong?
18TextSnake
From paper: TextSnake: A Flexible Representation for Detecting Text of Arbitrary Shapes
verified75.62018Paper ↗Code ↗Looks wrong?
19SLPR
From paper: Sliding Line Point Regression for Shape Robust Scene Text Detection
verified74.82018Paper ↗Code ↗Looks wrong?

Recall

Recall is the reported evaluation metric for scut-ctw1500. 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 Recallverifiedpapervendorcommunityunverified
RankModelTrustScoreYearLinksFix
01TextMamba
ResNet-50 backbone. Source: Table I, arxiv:2512.06657
verified88.52024Paper ↗Looks wrong?
02MixNet
From paper: MixNet: Toward Accurate Detection of Challenging Scene Text in the Wild
verified88.32023Paper ↗Code ↗Looks wrong?
03SRFormer (ResNet-50)
From paper: SRFormer: Text Detection Transformer with Incorporated Segmentation and Regression
verified87.72023Paper ↗Code ↗Looks wrong?
04DeepSolo (with pre-training)
Pre-trained on Synth150K+MLT17+IC13+IC15. Source: Table 7, arxiv:2305.19957
verified86.32022Paper ↗Code ↗Looks wrong?
05DPText-DETR (ResNet50)
From paper: DPText-DETR: Towards Better Scene Text Detection with Dynamic Points in Transformer
verified86.22022Paper ↗Code ↗Looks wrong?
06TextSnake
From paper: TextSnake: A Flexible Representation for Detecting Text of Arbitrary Shapes
verified85.32018Paper ↗Code ↗Looks wrong?
07TextFuseNet (ResNeXt-101)
From paper: TextFuseNet: Scene Text Detection with Richer Fused Features
verified85.12020Paper ↗Code ↗Looks wrong?
08I3CL + SSL
From paper: I3CL:Intra- and Inter-Instance Collaborative Learning for Arbitrary-shaped Scene Text Detection
verified84.62021Paper ↗Code ↗Looks wrong?
09EK-Net
ResNet-18 backbone. Source: Table 3, arxiv:2401.11704
verified83.742024Paper ↗Looks wrong?
10PAN
From paper: Mask R-CNN with Pyramid Attention Network for Scene Text Detection
verified83.22018Paper ↗Looks wrong?
11PAN-640
From paper: Efficient and Accurate Arbitrary-Shaped Text Detection with Pixel Aggregation Network
verified81.22019Paper ↗Code ↗Looks wrong?
12CRAFT
From paper: Character Region Awareness for Text Detection
verified81.12019Paper ↗Code ↗Looks wrong?
13FAST-B-640
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified80.92021Paper ↗Code ↗Looks wrong?
14FAST-B-512
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified80.22021Paper ↗Code ↗Looks wrong?
15PSENet-1s
From paper: Shape Robust Text Detection with Progressive Scale Expansion Network
verified79.892018Paper ↗Code ↗Source ↗Looks wrong?
16FAST-S-512
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified78.72021Paper ↗Code ↗Looks wrong?
17FAST-T-512
From paper: FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation
verified77.92021Paper ↗Code ↗Looks wrong?
18SLPR
From paper: Sliding Line Point Regression for Shape Robust Scene Text Detection
verified70.12018Paper ↗Code ↗Looks wrong?

F Measure Full Lexicon

F Measure Full Lexicon is the reported evaluation metric for scut-ctw1500. 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
01SPTS
From paper: SPTS: Single-Point Text Spotting
verified83.82021Paper ↗Code ↗Looks wrong?
02A3S
From paper: A3S: Adversarial learning of semantic representations for Scene-Text Spotting
verified82.32023Paper ↗Looks wrong?
03TESTR
From paper: Text Spotting Transformers
verified81.52022Paper ↗Code ↗Looks wrong?
04DeepSolo (ResNet-50)
From paper: DeepSolo++: Let Transformer Decoder with Explicit Points Solo for Multilingual Text Spotting
verified81.42023Paper ↗Code ↗Looks wrong?
05ABINet++
From paper: ABINet++: Autonomous, Bidirectional and Iterative Language Modeling for Scene Text Spotting
verified80.32022Paper ↗Code ↗Looks wrong?
06TPSNet
From paper: TPSNet: Reverse Thinking of Thin Plate Splines for Arbitrary Shape Scene Text Representation
verified79.22021Paper ↗Code ↗Looks wrong?
07MANGO
From paper: MANGO: A Mask Attention Guided One-Stage Scene Text Spotter
verified78.72020Paper ↗Code ↗Looks wrong?
08ABCNet v2
From paper: ABCNet v2: Adaptive Bezier-Curve Network for Real-time End-to-end Text Spotting
verified77.22021Paper ↗Code ↗Looks wrong?
09SwinTextSpotter
From paper: SwinTextSpotter: Scene Text Spotting via Better Synergy between Text Detection and Text Recognition
verified772022Paper ↗Code ↗Looks wrong?
10TextDragon
From paper: TextDragon: An End-to-End Framework for Arbitrary Shaped Text Spotting
verified72.42019Paper ↗Looks wrong?

F Measure No Lexicon

F Measure No Lexicon is the reported evaluation metric for scut-ctw1500. 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
01A3S
From paper: A3S: Adversarial learning of semantic representations for Scene-Text Spotting
verified64.42023Paper ↗Looks wrong?
02DeepSolo (ResNet-50)
From paper: DeepSolo++: Let Transformer Decoder with Explicit Points Solo for Multilingual Text Spotting
verified64.22023Paper ↗Code ↗Looks wrong?
03SPTS
From paper: SPTS: Single-Point Text Spotting
verified63.62021Paper ↗Code ↗Looks wrong?
04ABINet++
From paper: ABINet++: Autonomous, Bidirectional and Iterative Language Modeling for Scene Text Spotting
verified60.22022Paper ↗Code ↗Looks wrong?
05TPSNet
From paper: TPSNet: Reverse Thinking of Thin Plate Splines for Arbitrary Shape Scene Text Representation
verified59.72021Paper ↗Code ↗Looks wrong?
06MANGO
From paper: MANGO: A Mask Attention Guided One-Stage Scene Text Spotter
verified58.92020Paper ↗Code ↗Looks wrong?
07ABCNet v2
From paper: ABCNet v2: Adaptive Bezier-Curve Network for Real-time End-to-end Text Spotting
verified57.52021Paper ↗Code ↗Looks wrong?
08TextPerceptron
From paper: Text Perceptron: Towards End-to-End Arbitrary-Shaped Text Spotting
verified572020Paper ↗Code ↗Looks wrong?
09TESTR
From paper: Text Spotting Transformers
verified562022Paper ↗Code ↗Looks wrong?
10SwinTextSpotter
From paper: SwinTextSpotter: Scene Text Spotting via Better Synergy between Text Detection and Text Recognition
verified51.82022Paper ↗Code ↗Looks wrong?
11TextDragon
From paper: TextDragon: An End-to-End Framework for Arbitrary Shaped Text Spotting
verified39.72019Paper ↗Looks wrong?
§ 04 · Submit a result

Add to the leaderboard.

← Back to Document OCR