AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Vision-Language Pre-Training for Boosting Scene Text Detectors

Sibo Song, Jianqiang Wan, Zhibo Yang, Jun Tang, Wenqing Cheng, Xiang Bai, Cong Yao

CVPR 2022 · Vision-language and multimodal › Human and embodied modeling · approach: Vision-language pretraining · 42 citations (OpenAlex)

Check your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use VLPT-STD pre-training instead of STKMPSENet · Total-TextF-measure: 84.3 -> 86.1Improved
Use VLPT-STD pre-training instead of STKMEAST · MSRA-TD500F-measure: 80.0 -> 82.2Improved
Use VLPT-STD pre-training instead of STKMDB · MSRA-TD500F-measure: 85.9 -> 88.5Improved
Use the proposed FPN instead of the original FPN (ablation)VLPT-STD fine-tuned with PSENet · CTW1500F-measure: 82.7 -> 83.3Improved
Use the proposed FPN instead of the original FPN (ablation)VLPT-STD fine-tuned with EAST · ICDAR2015F-measure: 88.0 -> 88.3Improved
Add multi-head cross-attention rather than using only self-attention (ablation)VLPT-STD · CTW1500F-measure: 83.1 -> 83.3Improved
Add MLM alone rather than using SynthText detector pre-training (ablation)PSENet · ICDAR2015F-measure: 81.3 -> 84.5Improved
Add WIP alone rather than using SynthText detector pre-training (ablation)PSENet · ICDAR2015F-measure: 81.3 -> 83.1Improved
Add ITC alone rather than using SynthText detector pre-training (ablation)EAST · MSRA-TD500F-measure: 82.0 -> 79.3Hurt
Add WIP to ITC + MLM (ablation)PSENet · Total-TextF-measure: 85.6 -> 86.1Improved
Add WIP to ITC + MLM (ablation)EAST · ICDAR2015F-measure: 87.5 -> 88.3Improved
Add TextOCR to SynthText pre-training (ablation)PSENet with text supervision · ICDAR2015F-measure: 84.33 -> 85.07Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy