https://github.com/deepinsight/insightocr

MXNet OCR implementation. Including text recognition and detection.
https://github.com/deepinsight/insightocr

crnn mxnet ocr text-recognition

Last synced: 6 months ago
JSON representation

MXNet OCR implementation. Including text recognition and detection.

• Host: GitHub
• URL: https://github.com/deepinsight/insightocr
• Owner: deepinsight
• License: mit
• Created: 2018-06-21T11:21:13.000Z (over 7 years ago)
• Default Branch: master
• Last Pushed: 2018-09-17T09:08:48.000Z (about 7 years ago)
• Last Synced: 2024-11-03T10:32:48.547Z (11 months ago)
• Topics: crnn, mxnet, ocr, text-recognition
• Language: Python
• Homepage:
• Size: 38.2 MB
• Stars: 108
• Watchers: 17
• Forks: 43
• Open Issues: 5
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# insightocr

### Text Recognition Accuracy on Chinese dataset by [caffe-ocr](https://github.com/senlinuc/caffe_ocr)

| Network   | LSTM | 4x1 Pooling | Gray | Test Acc |

| --------- | ---- | ----------- | ---- | -------- |

| SimpleNet | N    | Y           | Y    | 99.37%   |

| SE-ResNet34 | N    | Y           | Y    | 99.73%   |

### Text Recognition Accuracy on [VGG_Text](http://www.robots.ox.ac.uk/~vgg/data/text/), on the subset of label size<=18

| Network   | LSTM | 4x1 Pooling | Gray | Test Acc |

| --------- | ---- | ----------- | ---- | -------- |

| SimpleNet | Y    | Y           | Y    | 87.17%  |

| SE-ResNet50-PReLU | Y    | Y           | Y    | 94.05%  |

| SE-ResNeXt101-PReLU | Y    | Y           | Y    | 94.38%  |