DeepErase

KO | EN

gitlite — search

#javascript #ai #llm #go #android #machine-learning #rust #deep-learning #macos #java #cli #vue #awesome-list #mcp

(★ 62)

A Unet based deeplearning model to line/box/spurious artifacts from text images. Unsupervised training.

DeepErase 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# DeepErase
- DeepErase is a U-net-like tensorflow sementic segmenation model removing artifacts (lines, boxes, spurious words) from text images extracted from documents. The cleansing of the artifacts enhances OCR performance over the image extractions.

### Authors
- [Yike Qi](yike.qi.cn@gmail.com), [Ronny Huang](mailto:wronnyhuang@gmail.com)

### Abstract
- We present a [method](https://github.com/yikeqicn/DeepErase/tree/master/src/DataFactory) to programmatically generate artiﬁcial text images with realistic-looking artifacts, and use them to train the U-net-like model in a totally unsupervised manner.
- The U-net-like model was trained in two modes:
 - [Standalone training](https://github.com/yikeqicn/DeepErase/tree/master/src/CleaningStandaloneModel): Optimize at Unet Segmentation loss only.
 - [Joint training with downstream Recognition model](https://github.com/yikeqicn/DeepErase/tree/master/src/CleaningRecognitionJointModel): Optimize at Unet Segmentation loss + recognition CTC loss. To balance image cleaning and recognition performance. ** The RNN CTC based [HTR model](https://github.com/wronnyhuang/htr) was used as the recognition model.
 
### Result
- Both validation pixel level segmentation accuracies were above 95%.
- Downstream recognition performances were evaluated on validation images and IRS extractions. The IRS extractions were extracted from [NIST sd02 tax forms](https://www.nist.gov/srd/nist-special-database-2), and were not used in model training. The word recognition accuracy were improved and beat the naive Hough cv2 cleaning method.

 <img src="https://github.com/yikeqicn/DeepErase/blob/master/Segmentation_Accuracy.JPG">


 <img src="https://github.com/yikeqicn/DeepErase/blob/master/Recognition_Validation_Set.JPG">


 <img src="https://github.com/yikeqicn/DeepErase/blob/master/Recognition_IRS_Set.JPG">

### Requirements
- python 3.5 or above
- tensorflow 1.12.0
- torch 0.4.1
- cv2 4.0.0

or simply
- docker pull wrhuang/default
- or docker pull jdegange/default
- with minor further pip install

DeepErase

DeepErase 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

DeepErase

DeepErase 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README