AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Sequence-to-Sequence Contrastive Learning for Text Recognition

Aviad Aberdam, Ron Litman, Shahar Tsiper, Oron Anschel, Ron Slossberg, Shai Mazor, R. Manmatha, Pietro Perona

CVPR 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining · 143 citations (OpenAlex)

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
replace whole-image SimCLR with sequence-level SeqCLRtext recognition representation learning · IAM, RIMES, CVL, IIT5K, IC03, and IC13 decoder evaluationaccuracyImproved
use window-to-instance rather than frame-to-instance mapping (ablation)SeqCLR · scene-text decoder evaluationaccuracy: 35.7 vs 3.0; 43.6 vs 4.6; 43.5 vs 4.9Improved
use window-to-instance rather than all-to-instance mapping (ablation)SeqCLR · scene-text decoder evaluation with CTCaccuracy: 35.7 vs 20.4; 43.6 vs 24.2; 43.5 vs 24.4Improved
use SeqCLR pretraining rather than supervised-only trainingtext recognizer · handwritten datasets with limited labeled dataaccuracy: IAM 5% CTC: 21.4 -> 31.2; RIMES 5% CTC: 35.9 -> 61.8Improved
use SeqCLR pretraining and fine-tune with all labels rather than supervised-only trainingtext recognizer · IAM and RIMES handwritten recognitionaccuracy: IAM attention: 77.8 -> 79.9; RIMES attention: 89.3 -> 92.5Improved
use SeqCLR pretraining rather than supervised-only trainingtext recognizer with attention decoder · scene-text datasetsaccuracyNo clear effect
use SeqCLR rather than the strongest listed word-level methodhandwritten text recognition · IAMWER: 22.2 -> 20.1Improved
use SeqCLR rather than the strongest listed word-level methodhandwritten text recognition · RIMESWER: 9.6 -> 7.6Improved
add a BiLSTM projection head (ablation)SeqCLR All-to-instance · RIMES representation evaluation, contextual features, CTCaccuracy: 55.4 -> 59.0Improved
add an MLP-per-frame projection head (ablation)SeqCLR Frame-to-instance · RIMES representation evaluation, contextual features, CTCaccuracy: 49.9 -> 57.5Improved
add a BiLSTM projection head (ablation)SeqCLR Window-to-instance · RIMES representation evaluation, contextual features, attentionaccuracy: 79.5 -> 75.9Hurt
add a BiLSTM projection head (ablation)SeqCLR All-to-instance · RIMES representation evaluation, visual features, CTCaccuracy: 2.0 -> 35.5Improved

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