craft_hw_ocr

KO | EN

gitlite — search

#python #hacktoberfest #react #machine-learning #docker #nodejs #macos #java #kubernetes #vue #pytorch #security #ios #open-source

(★ 28)

Recognition of handwritten text using CRAFT text detection and TrOCR

craft_hw_ocr Latest Version Download

Download Latest Version (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# craft_hw_ocr
Recognition of handwritten text using CRAFT text detection and TrOCR

## CRAFT

[CRAFT(Character Region Awareness for Text Detection)](https://arxiv.org/abs/1904.01941) is a way of doing OCR on images by exploring the region around the text. On high level the algorithm generates heatmap of the image and convolute upon them rather than on the image directly. Also for generating bounding boxes it will model respective thresholds with anchor points between individual character and individual words

## TrOCR

[TrOCR](https://huggingface.co/transformers/model_doc/trocr.html#) is essentially an encoder-decoder model, where encoder network creates an representation of the image using image encoding transformers models(ViT, DEiT) and the decoder network (language models) converts the processed repsentation into target strings.

## Implementation

```python
from craft_hw_ocr import OCR
import cv2
import numpy as np

img = OCR.load_image('/content/example_1.png')

# do the below step if your image is tilted by some angle else ignore
# img = OCR.process_image(img)

ocr_models = OCR.load_models()

img, results = OCR.detection(img, ocr_models[2])

bboxes, text = OCR.recoginition(img, results, ocr_models[0], ocr_models[1])

pilImage = OCR.visualize(img, results)

```

## Online Demo

[Huggingface Space](https://huggingface.co/spaces/vishnun/CRAFT-OCR)

## Credits

[craft-text-detector](https://github.com/fcakyon/craft-text-detector)

craft_hw_ocr

craft_hw_ocr Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

craft_hw_ocr

craft_hw_ocr Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

Download README