On Vocabulary Reliance in Scene Text Recognition筆記
簡介 目前很多sota的方法的結果在預測字典中的單詞有很好的的效果，但是遇到圖片中存在不存在字典中的單詞的效果就不是很好。作者把這種現象稱作爲vocabulary reliance.作者發現（1）這種現象是普遍存在的，或多或少都會有這種情況（2）基於attention的方法在預測詞典外的單詞效果不好，而使用視覺特徵的分割類的方法效果優秀。（3）內容建模和預測層相關。作者提出了一種結合基於att
相關文章
- 1. robust scene text recognition with automatic rectification
- 2. SRN: Towards Accurate Scene Text Recognition with Semantic Reasoning Networks ---論文閱讀筆記
- 3. Research on Pedestrian Attribute Recognition Based on Semantic Segmentation in Natural Scene
- 4. Robust Scene Text Recognition with Automatic Rectification 學習筆記
- 5. 【論文筆記】Robust Scene Text Recognition with Automatic Rectification
- 6. SEED: Semantics Enhanced Encoder-Decoder Framework for Scene Text Recognition ---論文閱讀筆記
- 7. 【text recognition算法】ESIR: End-to-end Scene Text Recognition via Iterative Image Rectification
- 8. 【text recognition算法】Scene Text Recognition from Two-Dimensional Perspective
- 9. Few-shot Learning for Named Entity Recognition in Medical Text筆記
- 10. （Paper）Robust Text Detection in Natural Scene Images
- 更多相關文章...
- • SQL IN 操作符 - SQL 教程
- • Swift for-in 循環 - Swift 教程
- • Tomcat學習筆記（史上最全tomcat學習筆記）
- • 使用阿里雲OSS+CDN部署前端頁面與加速靜態資源
相關標籤/搜索
每日一句
-
每一个你不满意的现在，都有一个你没有努力的曾经。
歡迎關注本站公眾號,獲取更多信息
相關文章
- 1. robust scene text recognition with automatic rectification
- 2. SRN: Towards Accurate Scene Text Recognition with Semantic Reasoning Networks ---論文閱讀筆記
- 3. Research on Pedestrian Attribute Recognition Based on Semantic Segmentation in Natural Scene
- 4. Robust Scene Text Recognition with Automatic Rectification 學習筆記
- 5. 【論文筆記】Robust Scene Text Recognition with Automatic Rectification
- 6. SEED: Semantics Enhanced Encoder-Decoder Framework for Scene Text Recognition ---論文閱讀筆記
- 7. 【text recognition算法】ESIR: End-to-end Scene Text Recognition via Iterative Image Rectification
- 8. 【text recognition算法】Scene Text Recognition from Two-Dimensional Perspective
- 9. Few-shot Learning for Named Entity Recognition in Medical Text筆記
- 10. （Paper）Robust Text Detection in Natural Scene Images