• File
  • Strongly Contextualised results
  • Weakly Contextualised results
  • Generic results
  • Evaluation
  • End-to-End recognition
  • Word Spotting

    • method: AIX Lab., LG Electronics 2021-02-05

    Authors: Wonju Lee and Joseph Lim

    Affiliation: AI Lab., Future Tech. Center, CTO, LG Electronics

    Email: wonju2.lee@lge.com

    Description: We reimplemented the end-to-end text spotting method combined the Mask TextSpotter with Deformable ConvNets. We tried to pre-train synthText images for character-level annotations and to fine-tune real world images like following papers for the generalization. We don't use any private data.

    Mask TextSpotter: An End-to-End TrainableNeural Network for Spotting Text with Arbitrary Shapes Minghui Liao, Pengyuan Lyu, Minghang He, Cong Yao, Wenhao Wu, Xiang Bai

    Deformable ConvNets v2: More Deformable, Better Results Xizhou Zhu, Han Hu, Stephen Lin, Jifeng Dai

    Source code

    method: TencentAILab2017-10-28

    Authors: Jingchao Zhou, Xing Ji, Zheng Zhou, Zhifeng Li

    Description: Our method consists of two networks: text detection network employed in task 2.1 and word recognition network employed in task 2.2. Two networks are integrated with cascade training to achieve superior performance.

    method: SRC-B-MachineLearningLab2017-07-11

    Authors: Shuli Yang, Yingying Jiang, Xiaobing Wang, Xiangyu Zhu, Pei Fu, Hua Wang, Zhenbo Luo

    Description: Samsung R&D Institute of China - Beijing. Machine Learning Lab. Our method is based on 2.1 and 2.3 (updated version). The paper is in preparation.

    Ranking Table

    Description Paper Source Code
    DateMethodRecallPrecisionHmean
    2021-02-05AIX Lab., LG Electronics 93.02%97.93%95.41%
    2017-10-28TencentAILab95.75%94.41%95.07%
    2017-07-11SRC-B-MachineLearningLab89.75%94.49%92.06%
    2018-01-22FOTS88.33%95.97%91.99%
    2016-08-23HUST_MCLAB87.68%95.83%91.57%
    2017-10-28MPTSys84.08%96.13%89.70%
    2016-03-17Adelaide_ConvLSTMs79.39%96.68%87.19%
    2016-03-10SRC-B-TextProcessingLab81.68%93.16%87.04%
    2015-04-03VGGMaxBBNet_09582.12%91.05%86.35%
    2015-04-03VGGMaxBBNet (055)82.99%89.63%86.18%
    2015-10-20Deep2Text II+72.08%94.56%81.81%
    2015-04-02StradVision-175.03%88.66%81.28%
    2015-04-02Deep2Text II-269.79%81.74%75.29%
    2015-04-01NJU Text (Version3)69.47%80.13%74.42%
    2015-04-01Deep2Text I66.74%83.95%74.36%
    2015-04-02MSER MRF61.40%84.53%71.13%
    2015-04-02Beam search CUNI (decoding for TextSpotter - no postprocessing)58.89%68.01%63.12%
    2015-04-02Beam search CUNI +S (decoding for TextSpotter - spell checked)15.27%92.72%26.22%
    2015-04-01Baseline-TextSpotter0.00%0.00%0.00%

    Ranking Graphic