Codesota · Benchmark · icdar-2017-mltHome/Leaderboards/Vision & Documents/Scene Text Detection/icdar-2017-mlt
Unknown

icdar-2017-mlt.

icdar-2017-mlt 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 icdar-2017-mlt.

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 ↗

Precision

Precision is the reported evaluation metric for icdar-2017-mlt. 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
01PMTD*
From paper: Pyramid Mask Text Detector
verified84.422019Paper ↗Code ↗Looks wrong?
02Corner Localization (single-scale)
From paper: Multi-Oriented Scene Text Detection via Corner Localization and Region Segmentation
verified83.82018Paper ↗Code ↗Looks wrong?
03TextBPN++ (ResNet-50+DCN)
TextBPN++ Res50-1s+DCN. Table IX. T-MM 2023.
verified83.742022Paper ↗Looks wrong?
04DBNet (ResNet-50+DCN)
DBNet (AAAI 2020) ResNet-50+DCN. Table IX of TextBPN++ (T-MM 2023).
verified83.12022Paper ↗Looks wrong?
05SBD
From paper: Exploring the Capacity of an Orderless Box Discretization Network for Multi-orientation Scene Text Detection
verified82.752019Paper ↗Code ↗Looks wrong?
06DBNet (ResNet-18+DCN)
DBNet (AAAI 2020) ResNet-18+DCN. Table IX of TextBPN++ (T-MM 2023).
verified81.92022Paper ↗Looks wrong?
07FOTS MS
From paper: FOTS: Fast Oriented Text Spotting with a Unified Network
verified81.862018Paper ↗Code ↗Looks wrong?
08CharNet H-88
From paper: Convolutional Character Networks
verified81.272019Paper ↗Code ↗Looks wrong?
09FOTS
From paper: FOTS: Fast Oriented Text Spotting with a Unified Network
verified80.952018Paper ↗Code ↗Looks wrong?
10CRAFT
From paper: Character Region Awareness for Text Detection
verified80.62019Paper ↗Code ↗Looks wrong?
11SPCNET
From paper: Scene Text Detection with Supervised Pyramid Context Network
verified80.62018Paper ↗Code ↗Looks wrong?
12TextBPN++ (ResNet-50)
TextBPN++ Res50-1s (no DCN). Table IX. T-MM 2023.
verified80.492022Paper ↗Looks wrong?
13PAN
From paper: Mask R-CNN with Pyramid Attention Network for Scene Text Detection
verified802018Paper ↗Looks wrong?
14GNNets
From paper: Geometry Normalization Networks for Accurate Scene Text Detection
verified79.632019Paper ↗Code ↗Looks wrong?
15CharNet R-50
From paper: Convolutional Character Networks
verified77.072019Paper ↗Code ↗Looks wrong?
16PSENet-1s
From paper: Shape Robust Text Detection with Progressive Scale Expansion Network
verified77.012018Paper ↗Code ↗Looks wrong?
17PSENet (ResNet-152)
From paper: Shape Robust Text Detection with Progressive Scale Expansion Network
verified75.352019Paper ↗Code ↗Looks wrong?
18Corner Localization (multi-scale)
From paper: Multi-Oriented Scene Text Detection via Corner Localization and Region Segmentation
verified74.32018Paper ↗Code ↗Looks wrong?

F Measure

F Measure is the reported evaluation metric for icdar-2017-mlt. 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
01PMTD*
From paper: Pyramid Mask Text Detector
verified80.132019Paper ↗Code ↗Looks wrong?
02SBD
From paper: Exploring the Capacity of an Orderless Box Discretization Network for Multi-orientation Scene Text Detection
verified79.472019Paper ↗Code ↗Looks wrong?
03TextBPN++ (ResNet-50+DCN)
TextBPN++ Res50-1s+DCN. Table IX / GitHub README. P=83.74, R=72.10, F=77.48. T-MM 2023.
verified77.482022Paper ↗Looks wrong?
04CharNet H-88
From paper: Convolutional Character Networks
verified75.772019Paper ↗Code ↗Looks wrong?
05DBNet (ResNet-50+DCN)
DBNet (AAAI 2020) ResNet-50+DCN. Table IX of TextBPN++ (T-MM 2023). P=83.1, R=67.9, F=74.7.
verified74.72022Paper ↗Looks wrong?
06GNNets
From paper: Geometry Normalization Networks for Accurate Scene Text Detection
verified74.542019Paper ↗Code ↗Looks wrong?
07PAN
From paper: Mask R-CNN with Pyramid Attention Network for Scene Text Detection
verified74.32018Paper ↗Looks wrong?
08SPCNET
From paper: Scene Text Detection with Supervised Pyramid Context Network
verified74.12018Paper ↗Code ↗Looks wrong?
09CharNet R-50
From paper: Convolutional Character Networks
verified73.422019Paper ↗Code ↗Looks wrong?
10PSENet-1s
From paper: Shape Robust Text Detection with Progressive Scale Expansion Network
verified72.452018Paper ↗Code ↗Looks wrong?
11Corner Localization (multi-scale)
From paper: Multi-Oriented Scene Text Detection via Corner Localization and Region Segmentation
verified72.42018Paper ↗Code ↗Looks wrong?
12TextBPN++ (ResNet-50)
TextBPN++ Res50-1s (no DCN). Table IX. P=80.49, R=65.67, F=72.33. T-MM 2023.
verified72.332022Paper ↗Looks wrong?
13PSENet (ResNet-152)
From paper: Shape Robust Text Detection with Progressive Scale Expansion Network
verified72.132019Paper ↗Code ↗Looks wrong?
14DBNet (ResNet-18+DCN)
DBNet (AAAI 2020) ResNet-18+DCN. Table IX of TextBPN++ (T-MM 2023). P=81.9, R=63.8, F=71.7.
verified71.72022Paper ↗Looks wrong?
15FOTS MS
From paper: FOTS: Fast Oriented Text Spotting with a Unified Network
verified70.752018Paper ↗Code ↗Looks wrong?
16FOTS
From paper: FOTS: Fast Oriented Text Spotting with a Unified Network
verified67.252018Paper ↗Code ↗Looks wrong?
17Corner Localization (single-scale)
From paper: Multi-Oriented Scene Text Detection via Corner Localization and Region Segmentation
verified66.82018Paper ↗Code ↗Looks wrong?

Recall

Recall is the reported evaluation metric for icdar-2017-mlt. 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
01SBD
From paper: Exploring the Capacity of an Orderless Box Discretization Network for Multi-orientation Scene Text Detection
verified76.442019Paper ↗Code ↗Looks wrong?
02PMTD*
From paper: Pyramid Mask Text Detector
verified76.252019Paper ↗Code ↗Looks wrong?
03TextBPN++ (ResNet-50+DCN)
TextBPN++ Res50-1s+DCN. Table IX. T-MM 2023.
verified72.12022Paper ↗Looks wrong?
04CharNet H-88
From paper: Convolutional Character Networks
verified70.972019Paper ↗Code ↗Looks wrong?
05Corner Localization (multi-scale)
From paper: Multi-Oriented Scene Text Detection via Corner Localization and Region Segmentation
verified70.62018Paper ↗Code ↗Looks wrong?
06CharNet R-50
From paper: Convolutional Character Networks
verified70.12019Paper ↗Code ↗Looks wrong?
07GNNets
From paper: Geometry Normalization Networks for Accurate Scene Text Detection
verified70.062019Paper ↗Code ↗Looks wrong?
08PAN
From paper: Mask R-CNN with Pyramid Attention Network for Scene Text Detection
verified69.82018Paper ↗Looks wrong?
09PSENet (ResNet-152)
From paper: Shape Robust Text Detection with Progressive Scale Expansion Network
verified69.182019Paper ↗Code ↗Looks wrong?
10SPCNET
From paper: Scene Text Detection with Supervised Pyramid Context Network
verified68.62018Paper ↗Code ↗Looks wrong?
11PSENet-1s
From paper: Shape Robust Text Detection with Progressive Scale Expansion Network
verified68.42018Paper ↗Code ↗Looks wrong?
12CRAFT
From paper: Character Region Awareness for Text Detection
verified68.22019Paper ↗Code ↗Looks wrong?
13DBNet (ResNet-50+DCN)
DBNet (AAAI 2020) ResNet-50+DCN. Table IX of TextBPN++ (T-MM 2023).
verified67.92022Paper ↗Looks wrong?
14TextBPN++ (ResNet-50)
TextBPN++ Res50-1s (no DCN). Table IX. T-MM 2023.
verified65.672022Paper ↗Looks wrong?
15DBNet (ResNet-18+DCN)
DBNet (AAAI 2020) ResNet-18+DCN. Table IX of TextBPN++ (T-MM 2023).
verified63.82022Paper ↗Looks wrong?
16FOTS MS
From paper: FOTS: Fast Oriented Text Spotting with a Unified Network
verified62.32018Paper ↗Code ↗Looks wrong?
17FOTS
From paper: FOTS: Fast Oriented Text Spotting with a Unified Network
verified57.512018Paper ↗Code ↗Looks wrong?
18Corner Localization (single-scale)
From paper: Multi-Oriented Scene Text Detection via Corner Localization and Region Segmentation
verified55.62018Paper ↗Code ↗Looks wrong?

H Mean

H Mean is the reported evaluation metric for icdar-2017-mlt. 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 H Meanverifiedpapervendorcommunityunverified
RankModelTrustScoreYearLinksFix
01CRAFT
From paper: Character Region Awareness for Text Detection
verified73.92019Paper ↗Code ↗Looks wrong?
§ 04 · Submit a result

Add to the leaderboard.

← Back to Scene Text Detection