Research map / Detection and segmentation / CNN features
An End-to-End Trainable Neural Network for Image-Based Sequence Recognition and Its Application to Scene Text Recognition
Baoguang Shi, Xiang Bai, Cong Yao
Read the paperCheck 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 changed | Where | Measured | Effect |
|---|---|---|---|
| use CRNN instead of the 90k-dictionary classifier | scene text recognizer · IIIT5k (lexicon 50) | accuracy (%): 97.1 -> 97.6 | Improved |
| use CRNN instead of the 90k-dictionary classifier | scene text recognizer · SVT (lexicon 50) | accuracy (%): 95.4 -> 96.4 | Improved |
| use CRNN instead of the 90k-dictionary classifier | scene text recognizer · IC03 (Full lexicon) | accuracy (%): 98.6 -> 97.6 | Hurt |
| use CRNN instead of the 90k-dictionary classifier without lexicon | scene text recognizer · IC03 (no lexicon) | accuracy (%): 93.1 -> 89.4 | Hurt |
| use CRNN instead of the unconstrained 304M model without lexicon | scene text recognizer · SVT (no lexicon) | accuracy (%): 71.7 -> 80.8 | Improved |
| use CRNN with 8.3M parameters instead of 304M-parameter model | scene text recognizer · model size | parameters: 304M -> 8.3M | Improved |
| use CRNN instead of Capella Scan on real-world scores | pitch recognition · Pitch recognition (Real-World) | fragment accuracy (%): 43.5% -> 84.0% | Improved |
| use CRNN instead of PhotoScore on clean scores | pitch recognition · Pitch recognition (Clean) | fragment accuracy (%): 55.0% -> 74.6% | Improved |
| use CRNN instead of Capella Scan on synthesized scores | pitch recognition · Pitch recognition (Synthesized) | average edit distance: 2.31 -> 0.30 | Improved |
| increase lexicon search radius delta (ablation) | lexicon-based transcription · IC03 (lexicon 50k) | accuracy | Improved |
Closest papers by meaning
- SEE: Towards Semi-Supervised End-to-End Scene Text Recognition (AAAI 2018 · similarity 0.87)
- Towards End-To-End Text Spotting With Convolutional Recurrent Neural Networks (ICCV 2017 · similarity 0.87)
- Mask TextSpotter: An End-to-End Trainable Neural Network for Spotting Text with Arbitrary Shapes (ECCV 2018 · similarity 0.87)
- Reading Scene Text in Deep Convolutional Sequences (AAAI 2016 · similarity 0.87)
- OTE: Exploring Accurate Scene Text Recognition Using One Token (CVPR 2024 · similarity 0.87)
- Image as a Language: Revisiting Scene Text Recognition via Balanced, Unified and Synchronized Vision-Language Reasoning Network (AAAI 2024 · similarity 0.86)
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.
