A brief introduction to OCR (Optical character recognition)
3 likes21,069 views
The document discusses optical character recognition (OCR), explaining its purpose of reading various forms of text from images and emphasizing its challenges compared to object detection. It highlights difficulties such as variations in font, size, and language, as well as the need for advanced machine learning techniques. The content also suggests foundational knowledge required in machine learning and neural networks to understand OCR applications.
![Text Localization Text
Localization
Text
Recognition
์ดํ์, https://www.slideshare.net/deview/111-ai
regression-based
(like object detection)
end-to-end
[Textboxes, Liao et al., AAAI2017] [PixelLink, Deng et al., AAAI2018]
classification-based
(like semantic segmentation)
[FOTS, Liu et al., CVPR2018]
simultaneous local+recog
# of
papers
training unstable stable](https://image.slidesharecdn.com/42ocr-190214054348/85/A-brief-introduction-to-OCR-Optical-character-recognition-7-320.jpg)
A brief introduction to OCR (Optical character recognition)
- 1. Terry Taewoong Um fb.com/deeplearningtalk fb.com/terryum ์ฌ์ง ์ ๊ธ์๋ฅผ ์ฝ์ด์ฃผ๋ Optical Character Recognition (OCR) 42
- 2. What is OCR? โข Optical Character Recognition (OCR) Reading typed/printed/handwritten characters from image sources Speech Recognition
- 3. What is OCR? โข Optical Character Recognition (OCR) Reading typed/printed/handwritten characters from image sources OCR
- 4. Why OCR? characters in the computer characters in the physical world A
- 5. Why OCR? characters in the computer characters in the physical world Difficult because of the large variations! (font, size, shape, location, noise, ...)
- 6. OCR vs Object detection Text Localization Text Recognition โข OCR โข Object detection Object Localization Object Recognition Detect the bounding boxes that enclose text Read it โข OCR is more challenging than object detection due to - various aspect (W:H) ratio - large distortions - confusion w/ textures (โIโ, โTโ) - few pretrained models- high density - various languages
- 7. Text Localization Text Localization Text Recognition ์ดํ์, https://www.slideshare.net/deview/111-ai regression-based (like object detection) end-to-end [Textboxes, Liao et al., AAAI2017] [PixelLink, Deng et al., AAAI2018] classification-based (like semantic segmentation) [FOTS, Liu et al., CVPR2018] simultaneous local+recog # of papers training unstable stable
- 8. Text Recognition Text Localization Text Recognition Connectionist Temporal Classification r EOSpt i t pi<GO> r Attention # of papers speed rarely used accuracy
- 9. OCR + Translation = SmartLens Text Localization Text Recognition Machine translation โข What you need to know is - Machine learning basics - Neural network basics - Convolutional Neural Networks (+ advanced topics) - Recurrent Neural Networks (+ advanced topics)